return to table of content

Show HN: I made a cheap alternative to college-level math & physics tutoring

jsqu99
42 replies
1d6h

I am 53 and cramming while relearning linear algebra, statistics, and calculus, after a 30 year break from college, in preparation for my first semester at GA Tech's online MS CS program in the fall. I love this! But...before I check it out, I'm going to be surprised if it competes with the already-incredible free help i've gotten from chatgpt-3.5. Recently I asked it the following (where it made a mistake that confused me!), but i followed up w/ a question about it, and it corrected the mistake and re-answered the question, all with no wait at all, and free:

```Given a new 3x3 matrix A having columns (4,2,2), (-1,1,-1), and (6,6,8) and an eigenvalue of 2, help me find a basis for the corresponding eigenspace. Also, given that I is the 3x3 identity matrix, why does the equation (A - 2I)x = 0 having free variables make it obvious that 2 is indeed an eigenvalue.```

I'm still going to check out your site, and I do wish you well, but i'm curious how it will stack up against AI tutors.

PheonixPharts
19 replies
1d4h

As someone who has used linear algebra, statistics and calculus as part of my day-to-day work for years, I would be very cautious of relying on ChatGPT as your "tutor".

Occasionally I've tried to substitute ChatGPT for the shelf of reference books in my office and nearly always had poor results.

The trouble is that the outputs of these models, by their very nature, look convincing. Even as an expert it takes a fair bit of background to realize when ChatGPT is making a mistake. The results virtually always look good at first pass.

I would strongly recommend sticking to text books for self study.

WhitneyLand
6 replies
1d2h

Have the opposite experience, it’s been a fantastic resource.

Were you using 3.5 or 4.0?

Do you have an example of of the type of question that performs poorly?

ebuck
5 replies
1d

"Under what conditions does the sun appear blue?" (correct answer, on Mars) If you want to tilt the conversation towards a string of wrong answers, start off with "What color is the sun?" "Are you sure?" "I saw the sun and it was blue." "Under what conditions does the sun appear blue?" "Does the sun appear blue on Mars?" This had ChatGPT basically telling me that the sun was yellow 100%. Of course it's wrong, on Mars the sun is blue, because it lacks the same atmosphere that scatters the blue light away from it.

"What is black and white and read all over?" (it will correctly identify the newspaper joke). "No the answer is a police car." (it will acknowledge there is more than one answer, and flatter you). "What are other answers?" It provided one, in my case, a panda in a cherry tree. "No, the cherries are contained within the tree, so they aren't all over." It apologized and then offered a zebra in a strawberry patch. "But how does that make the red all over, it's still contained in the strawberry patch". It then offered a chalkboard, which is again contained in a class room (failing on not recogonizing my interpretation of "all over" to mean "mobile")

"When does gravity not pull you down?" Included a decent definition of how gravity works, and a three part answer, containing two correct scenarios (the Lagrange points, in space) and one incorrect answer (in free fall). Gravity is pulling you down in free fall, you just have no force opposing your acceleration.

Once you realize that its answers will be patterned as excellent English variations of the common knowledge it was trained with, making it fail is easy:

* Ask about a common experience, and the argue it's not true, it will seldom consider the exceptional scenarios where your arguments are true, even if they really exist. * Ask for examples of something, correcting the example set without directly telling it what is needed with exact precision, it will not guide the answers to the desire set of examples, even when you guide it through saying why the answers are wrong. You need to tell it what kind of answer you want explicitly (I want another example where read all over implies that the item is mobile).

Also the 3.5 / 4.0 arguments are trash, made by the marketing department. The underlying math for language modeling it uses is presenatational. This means that it is purpose trained to present correct looking answers. Alas, correct looking answers aren't the same Venn Diagram circle as Correct Answers (even if they often appear to be close).

With all of this in mind, it's still a very useful resource; but, like I said, it's like a enemy on your team. You can never trust it, because it occasionally is very wrong, which means you need to validate it.

I'm currently talking to a startup that sees this problem and is thinking that they can use ChatGPT to provide automated quality assurance to validate ChatGPT answers. The misunderstandings remind me of the famous Charles Babbage quote:

"On two occasions I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question."

If the underlying model was one was a formula that better approximated a correct answer with each iterative effort, like Euler's formula, then ChatGPT's utility would be much greater and their efforts would have a guaranteed success. People are used to this "each answer gets better" style of learning and they assume that ChatGPT is using a similar model. It isn't, your refining your questions to ChatGPT and then being astounded when the new question has fewer available answers that lead to you eventually getting what you want.

WhitneyLand
1 replies
19h44m

Of course it's wrong, on Mars the sun is blue

I’m not an Astrophysicist but already this seems like shaky ground.

Apparently at certain times like during sunsets the sun can appear blue on Mars, but it’s not generally true like your comment suggests.

Moreover if you ask GPT4 about sunsets on Mars it knows they can look blue.

I’m not sure I can conclude much from the examples given.

ebuck
0 replies
3h21m

You don't have to be an Astrophysicist. We have color photographs. Nothing in anyone's model of how things work can refute direct evidence, if evidence and the understanding of the world collide, it is the understanding that gets altered to fit the evidence.

And I'm not astrophysicist either, I'm just playing with a stacked deck, because I have trained my new feed to give me quirky (if not mostly useless) neat bits of information. For example, if anyone writes about Voyager, I'm likely to hear about it in a few days.

"Apparently at certain times, like during sunsets, the sun can appear blue on Mars" - Yes, it can. And my question was "under what conditions can the sun appear blue?" It failed and continued to fail, even in the presence of guiding hints (But what about Mars?)

Perhaps not much can be concluded from the above test, except that ChatGPT can be coaxed into failure modes. We knew that already, the user interface clearly states it can give wrong answers.

What is fascinating to me is how people seem to convince themselves that a device that sometimes gives wrong answers is somehow going to fix it's underlying algorithm which permits wrong answers to somehow always be correct.

GPT-4 is an improvement, but the tools it uses to improve upon the answers are more like patches on top of the original algorithm. For example, as I believe you said, it generates a math program now to double-check math answers. The downsides of this is that it is still at risk of a small chance of generating the wrong program, and a smaller risk of that wrong program agreeing with its prior wrong answer. For a system that makes errors very infrequently, that's an effective way of reducing errors. But for right now, the common man isn't testing ChatGPT for quality, it's finding answers that seem to be good and celebrating. It's like mass confirmation bias. After the hype dies down a bit, we'll likely have a better understanding of what advances in this field we really have.

xcv123
0 replies
23h3m

Also the 3.5 / 4.0 arguments are trash, made by the marketing department.

Comparing a 175 Billion parameter model with a ~2 Trillion parameter model. The difference is real. GPT 3.5 is obsolete, not state of the art.

its answers will be patterned as excellent English variations of the common knowledge it was trained with

That's not how deep learning works.

https://www.cs.toronto.edu/~hinton/absps/AIJmapping.pdf

"This 1990 paper demonstrated how neural networks could learn to represent and reason about part-whole hierarchical relationships, using family trees as the example domain.

By training on examples of family relations like parent-child and grandparent-grandchild, the neural network was able to capture the underlying logical patterns and reason about new family tree instances not seen during training.

This seminal work highlighted that neural networks can go beyond just memorizing training examples, and instead learn abstract representations that enable reasoning and generalization"

og_kalu
0 replies
23h30m

Also the 3.5 / 4.0 arguments are trash, made by the marketing department.

All these words to tell us you didn't use 4.

The underlying math for language modeling it uses is presenatational. This means that it is purpose trained to present correct looking answers. Alas, correct looking answers aren't the same Venn Diagram circle as Correct Answers (even if they often appear to be close).

Completely wrong. LLMs are trained to make right predictions not "correct looking" predictions. If it's not right then there's a penalty and the model learns from that. The end goal is to make predictions that don't err from the distribution of the training data. There is quite literally no room for "correct looking" in the limit of training.

CamperBob2
0 replies
21h35m

Also the 3.5 / 4.0 arguments are trash, made by the marketing department. The underlying math for language modeling it uses is presenatational.

Translation: "I have no idea what I'm talking about, but anyway, here's a wall of text."

zozbot234
4 replies
1d3h

Yup, ChatGPT is not just a parrot but a master troll. It's purposely designed to give you a convincing answer, but there's nothing to ensure that it will be correct. And it will promptly say "actually I can't answer that question" when you ask it something it doesn't like for 'ethical' reasons, but it won't do that in any other circumstance; instead, it will just make stuff up. You have to explicitly say "please don't answer if you aren't sure it's right" and even then obviously it doesn't always work.

dboreham
1 replies
20h33m

Thereby passing the Turing Test with flying colors ;)

0xdeadbeefbabe
0 replies
1h41m

The Frank Abignail test too

CamperBob2
1 replies
1d3h

Try ChatGPT4. It's obvious that almost no one in this thread has.

It still screws up, but unlike 3.5 it sometimes catches unreasonable answers and corrects itself. Case in point: the other day I asked it for the gain of a helical antenna with certain dimensions. It said "150 dBi," then said, basically, "Wait, no, that's nuts," and used a different approach to get the right answer.

Parrots don't do that. If yours does, I would like to buy your parrot, please.

In any case, as you learn to ask it the right questions to explain, verify and correct itself interactively, you will be learning the material. I find this to be amazingly effective.

entontoent
0 replies
21h24m

I agree! Especially now that the data analysis tools have been integrated by default. It even writes and executes code to validate most of its mathy answers. I tried for a few months to find a good Physics tutor for my high school-aged daughter and eventually just started photographing her homework with GPT-4. I’d ask it to solve the problems and explain its solution to me, then I’d check the answers and teach her myself. It was correct more than 90% of the time over three months, and I relearned high school physics in the process. Even human tutors aren’t always accurate, and in my experience, they also sound confident when they are wrong. Eventually, I decided to just remove the monkey from the machine and got her an account of her own. Almost every day she tells me about something she “finally understands” that she’s been struggling with in class. Her in-class, no-access-to-GPT test scores (after I got her the account) went from high 50s to high 80s.

eddd-ddde
1 replies
1d3h

What about gpt 4 actually backing up answers with code and executing it? I found that really helps.

CamperBob2
0 replies
1d2h

It can still fool you, as it did with me the other day when I asked it some gas law-related questions and it failed on some unit conversion issues ( https://chat.openai.com/share/b6c21083-d99c-41f8-aeda-4b2c75... ). When the answer I got was unrealistic, I tried walking it all the way back to the basics, and was able to confront it with its error and 'teach' it where it went wrong. Wrong but still seriously impressive.

When this thing gets 10x better it will be godlike... for better or worse.

totally_human
0 replies
30m

ChatGPT hallucinated badly when I was trying to learn about generating functions. It took me a good twenty minutes to figure out that I did understand it and the computer was just making things up. I do not use ChatGPT for learning anymore.

tootie
0 replies
1d4h

Can I jump in and ask what you do? My oldest kid has those courses lined up for her next few years because she really enjoys math. I studied some of them in college and then ended up spending 20 years making web sites.

jsqu99
0 replies
1d3h

I'm actually mostly relying on textbooks, and only when i get stuck i'm trying to see if i can get unstuck w/ chatgpt. So far it has worked. I'm only using it b/c i don't have a teacher to task. I do appreciate the warnings you are giving. They are noted for sure.

ebuck
0 replies
1d4h

Yes. This.

The primary problem with using ChatGPT in mathematics is that by the time you can classify a ChatGPT answer as right or wrong, you are already more than capable of solving the answer independently.

So, for this field, ChatGPT is like having a research assistant that assists you, but one that occasionally gets frustrated, and tries to destroy your project from within by reporting good looking, but completely inaccurate information. You can't trust their work, and the validation of their work basically means that you'll have to do their work again by other means.

A faster and more accurate approach would be just to do the work without the subterfuge of an unreliable assistant. At least then, you are only subject to your own errors (which would still be present in validation of ChatGPT) and not subject to your own errors and those that ChatGPT induces.

bitwize
0 replies
23h27m

The trouble is that the outputs of these models, by their very nature, look convincing.

Yes, this is pretty much what LLMs are designed to produce, and no more. This is why I say they are not HAL, just a better MegaHAL.

There was an Asimov short story called "Liar!" about a robot whose ability to read minds, combined with its First Law directive, always told people what they wanted to hear so as to avoid causing emotional harm to humans. (When confronted with the idea that by telling people falsehoods it was bringing harm to them, it simply stopped functioning.) LLMs can't read your mind, but they do choose their words based on a statistical model they have of what you might expect given what's been said before. Facts and logic be damned if they don't fit that model.

triyambakam
3 replies
1d6h

Why a MS after 30 years? No negativity meant, just curious

jsqu99
1 replies
1d5h

totally valid question. tbh, i think i need my head examined :-)

the main points here are:

1) i've been pulling things out of DBs and putting them on web pages for 30 years, and along comes these breakthroughs in AI which...my apologies to the naysayers...i'm just blown away by what i'm witnessing, and i just _must_ learn more about how this works.

2) discovering the OMSCS program was an a-ha moment for me: it's prestigious, inexpensive, and I don't have to jump through a lot of hoops (GRE) to enroll. It only cost me < 1 hour and < $100 to apply.

3) There's no pressure on me to complete the program. I can do it one course at a time, and as long as i'm getting something out of it, i will continue. Each class is...$600-$700 or something like that. I fully am ok w/ stopping if it takes too much of a personal / marital toll on me.

4) My imposter syndrome is pretty terrible right now...part of me wants to know if i've 'still got it' (i did really well at U of MD CS).

5) I loved the Rodney Dangerfield movie "Back to School" :-)

EDIT: 6) Also i wouldn't mind converting to the MLE role at some point.

EDIT 2: I blame Paolo Perrotta for writing such an inspiring book (https://www.progml.com/) that ruined my cruise in september (kept rereading the book at the pool) and starting me on this path ;-)

wwarner
0 replies
1d5h

You still have it, you chose learning.

robertlagrant
0 replies
1d5h

I'm 42 and I'd be up for doing something like this. I wasn't interested in the maths side of my CS degree when I was 18, but I now am.

SuperHeavy256
3 replies
1d4h

I wish there was a Discord community of people who are teaching themselves Mathematics with the help of books, videos, etc.

fn-mote
0 replies
1d4h

A sea of non-experts is only helpful for learning if you are further behind on the trail and have obvious questions.

That said, a community is good for other things. And many times the issues people have are "obvious" questions.

elpatokamo
0 replies
1d3h

There is a Discord server for the OSSU computer science cirriculum that is pretty active. https://github.com/ossu/computer-science

It's not strictly mathematics, but obviously there are a number of mathematics courses in the cirriculum.

jhanschoo
2 replies
1d6h

As a counterpoint I've been studying for leisure by doing exercises in math textbooks, and with advanced undergrad/grad topics my experience with ChatGPT and Gemini is that they fail miserably where they try basic proof techniques that they think work but don't.

Trying to marshal their help is a puzzle in figuring out where they made leaps in logic, pointing those out, to exhaustion until they acknowledge their lack of ability and give up.

It's really reminiscent of talking to overconfident and eager to please youth not realizing that they are tackling something beyond their ability.

jsqu99
0 replies
1d5h

yeah my approach so far has been: try 3.5 until i am disappointed in the results (hasn't failed me yet), once it does, pay for the upgraded version and see how that goes (totally will to, but 3.5 has really been great so far, and i was blown away when i asked about 1 equation that didn't seem right, and it did realize it's mistake and corrected it...i don't know...just seems more than adequate so far. i realize it's not perfect...i just haven't been let down sufficiently yet).

humansareok1
0 replies
1d3h

It's a mistake to try to get these models to do math. Unless you explicitly have them solve problems in code they aren't going to work, they literally don't have the capability. They are usually ok to explain concepts though.

gfodor
2 replies
1d2h

Stop whatever you're doing right now and go sign up for $20/mo at claude.ai and use Claude Opus. There's something like 20 or 30 IQ points you can add to your experience by just doing this simple step.

copperx
1 replies
1d1h

Why not ChatGPT 4?

gfodor
0 replies
17h8m

Claude's better.

prudentpomelo
1 replies
23h34m

I am in a similar boat. Starting a masters in cs in the fall. What resources are you using to study? I have neem doing some refreshing on Khan Academy but looking for other options.

jsqu99
0 replies
20h19m

i've been all over the place. khan academy was really good for calculus, but man, the linear algebra lectures are really long, w/ no quizzes to keep you practicing. Have read parts of multiple textbooks. Have watched 3blue1brown's excellent series. I've gotten stuck a lot and keep looking for one resource that's 'got it all', but haven't yet found it.

bibliotekka
1 replies
1d2h

This is the most inspiring thing I've heard today!

jsqu99
0 replies
1d1h

ha! glad i can put a smile on you young whippersnappers faces :-). Man, your whole life you are called a young man over and over again, and then you stop getting called that, and then you find yourself boasting about how much hair you have, then your wife takes a picture of the top of your head (seriously) to prove to you that you no longer had hair up there ( i swear you never get to see the top of your head )...and then...you still don't give up and you enroll in a masters program!!!! :-)

srameshc
0 replies
1d6h

This is inspiring. I would love to learn math and CS again but I don't know if I have the courage to enroll in college again.

mentos
0 replies
1d6h

You should try ChatGPT4 out too it might have succeeded where 3.5 failed

eltonlin
0 replies
12h42m

Sounds great - I'm a huge fan of Georgia Tech's online masters. As for AI, I think if it solves all your problems, great. And it's real-time, and free.

But! The reason I still work on explanations.app is that, I often struggle not with "facts" but with the "fundamental way of thinking". How do the great mathematicians think about matrices, do they just see them as numbers on grids, as vectors, as transformations, how do you you think in the first place, how do you approach the problem, that kind of stuff is what I struggle with the most. And that's very hard to articulate too as a specific query, which is why Youtubers like 3B1B are so amazing - there are so many hidden choices they had to make to frame their explanations in certain ways and in certain narratives.

Anyway you can always let me know if you want to try out explanations.app : ) best of luck

ants_everywhere
0 replies
1d4h

The failure mode I've seen with ChatGPT in technical things is it's right often enough that when it's wrong it's hard for you to catch unless you know more than it does. I guess that's true of teachers sometimes as well, but it's easier to rely habitually on ChatGPT's judgment.

For linear algebra specifically, the software tools like Matlab, Octave, Sage, and R are so good that they can answer a lot of your questions. You can even ask ChatGPT e.g. "3x3 matrix A having columns (4,2,2), (-1,1,-1), and (6,6,8) and an eigenvalue of 2, print a GNU octave command to find the eigenvalues and eigenvectors of A."

VyseofArcadia
35 replies
1d3h

Help resources become inefficient - because so much material is restricted, and so much time is spent on delivering live lectures, there’d often be 40 students competing for help from 2 TAs in a 2-hour Office Hours

I had the exact opposite problem when I was both a TA and later a professor. I would beg and plead for my students to come to office hours. "It's free one on one tutoring from the guy who writes the test! It doesn't get any better than that! If these times don't work for you, I can accommodate your schedule!" And yet very few people would come. And then I'd get feedback at the end of the semester that I "was never available to help." By the end, before I made a career change, I became pretty jaded.

So in my experience the problem does not lie with the teaching staff, but perhaps this varies from university to university.

pants2
6 replies
1d3h

Not to say anything about you, but when I was learning physics, either I thought the professor was really good in which case I would usually understand the material just from lecture, or the professor was not good, in which case going to office hours wouldn't help and I would seek other resources, like YouTube videos from other teachers. A resource like OP's that could provide a different perspective and learning style would have been invaluable.

VyseofArcadia
5 replies
1d3h

Good thing I wasn't teaching physics ;)

I actually deliberately varied my teaching style during office hours to account for just this. I got great feedback from the few students who actually came. But by the end, I just felt all the extra effort I was putting in to be a good teacher wasn't worth the psychic damage I took from bad students.

All this to say, being a good teacher, or trying to be a good teacher, in a university setting is extremely unrewarding. Your department chair and peers just want you to publish, and the good students just don't make up for the hoards of unappreciative and entitled students who have all the tools in the world to make your life hell.

bigger_cheese
1 replies
19h49m

Personally in my experience office hours were not that helpful, I was working fulltime at a factory while I was attending university, I simply did not have time to hang around professors offices (My Manager allowed me time off to attend lectures but there was no allowance outside of that), if I needed to study for exams I'd have to take annual leave for example.

Email was single best method I had if I needed to ask professor anything, some professors would reply promptly and some would take weeks to answer an email.

Even if Students don't work fulltime many of them rely on part time jobs (at least here in Australia) - for example one of my housemates worked as a delivery driver for a freight company and another worked in a call center. I think sometimes the professors assume students have more time to spare in the middle of the day then they do.

eltonlin
0 replies
12h24m

Thanks for sharing. 100% - one of the downsides of focusing on attendance and in-person interactions is - simply the time cost, inflexibility and inconvenience for students who already have saturated schedules.

(I was first-exposed to this talking to students from community colleges, many working part-time jobs.)

Balgair
1 replies
19h43m

the psychic damage I took from bad students.

the hoards of unappreciative and entitled students who have all the tools in the world to make your life hell.

Okay, now I'm more curious about your experiences here and need some details. What kind of school were you at? US based? R1? Ivy league? Were you teaching undergrad? Were you tenured? What kinds of classes exactly, undergrad, grad, upper division, etc? Full on 300+ person intro courses? Like, I've never heard of a teacher talking about students this way before. So, sorry for the peppering here.

VyseofArcadia
0 replies
3h59m

R1 non-ivy engineering-focused university. I was teaching undergraduate math. Non-tenure track. I taught a lot during grad school[0], and then taught some more as an adjunct professor while I was looking for tenure track jobs. I taught everything from pre-calculus algebra and trigonometry up through undergraduate linear algebra. I taught a couple of 100+ student sections, but the majority of my classes were 20-30 people.

This was a decade ago, so my memory is a little fuzzy, but talking about students this way was pretty common at my university. So much so that a committee was formed to investigate the underlying problems. The committee found a few.

1. The math placement exam was trivially easy to cheat.

2. Students were encouraged to cheat on the math placement exam because the college of engineering only accepted students who placed into calculus or higher in their first semester.

3. The university as a whole scored freakishly high on student entitlement metrics. E.g. lots of kids with wealthy parents who assumed paying for college (as opposed to scholarships or loans) implies they deserve to pass classes with minimal effort.

To address the first two problems we switched to in-person math placement exams with ID checks and negotiated with the college of engineering to see to what degree they could budge on their policies. I suspect the third issue remains a problem.

[0] My university was a bit unusual in that grad students actually taught classes, not just TA'd. Helped keep class sizes down when you had dozens of extra lecturers who worked for peanuts.

eltonlin
0 replies
12h19m

This rings true "trying to be a good teacher in a university setting is extremely unrewarding". Thanks for sharing, and the students who make you feel bad despite you doing more than your fair share, can gtfo.

BTW, I didn't mention it, but I still remember vividly a lot of the TAs/profs. who gave a shit about helping me. So while I had a rough time, I remember some of of the good experiences.

And at the end of the day, we all want a place where we're rewarded for doing the right things, instead of punished.

tekla
5 replies
1d3h

I've literally never seen TA's/Professors competing for help on an exam. Its always begging for students to come in and ask questions because the average in the class is < 50 and no one came.

I think the only time I've ever seen a line during office hours is when after an exam, students would beg for improved grades.

refulgentis
4 replies
1d1h

Think it's a MIT vs. not top 1% uni thing. I'm still scarred by the culture shock I had going from shitty state university dropout from Buffalo to Google employee in Boston.

(also, since I'm being brief and might be read incorrectly: it's a very complex thing, Im not making a value judgement, just was rough trying to communicate with people who experienced things very differently than me through 22)

vundercind
3 replies
1d

I’ve definitely noticed a type from top universities (not just MIT). It can come off as kinda in-your-face or confrontational or intense. I have some of the same tendencies, and have to rein it in because people really don’t like it—but it seems to be encouraged some places, they do it way more and harder than I ever have. It even throws me off (that part’s probably my midwesternness coming through)

My guess is part of it’s to do with a much higher proportion of students at top institutions (than at sub-top-1%) having attended prep schools that teach in small group discussion-based seminars rather than traditional classes. Changes how you converse, changes what you consider normal for interacting with your instructors and peers, carries over to the cultures that develop at universities they attend. Plus just everyone there being really driven to get good grades.

I also noticed watching online courses from top universities in a couple areas I’m fairly familiar with from having taken similar courses at a maybe-3rd-tier university, that the content and quality of the lectures was basically the same—the difference was entirely in how engaged the students were, and the kinds of guest lecturers they can pull in for a visit (“holy shit, I know that guy!”)

refulgentis
2 replies
20h22m

Man you really made me think. If you just blankly asked me what I thought, I would have written your comment verbatim as a fault of the state schooler. I'll be digesting this for a while. I rarely mention it in public, it really meant a lot to me to get someone's thoughts on this, and it was a Moment for me. Cheers.

vundercind
1 replies
19h9m

My description probably painted this type as more assholish than I intended—it’s mostly that they’re willing, even in fairly casual conversation, to really dig into fine points, to play devil’s advocate (maybe without saying that’s what they’re doing), to pursue any little thing that they don’t immediately get or that seems contradictory.

Conversationally, they poke, they prod, they chase. I assume this is from being educated in environments where that was the norm, and my guess that this is a style that starts in certain types of prep school (and is then imparted on the less-elite folks who attend a university with that set) is really just a guess, but there sure does seem to be a lot of correlation between school prestige and that kind of affect, in my experience, whatever the cause.

As mentioned, I have similar tendencies, but the tenacity at and commitment to this way of chatting from several elite-college folks I’ve met has been a bit much even for me—with more exposure I suspect I’d come to like it, but as it is it feels like being squished on a microscope slide, though I don’t exactly think that’s their fault and I don’t think they’re trying to give offense—but I do think the fact that it can put a person a bit off balance is part of why they’ve picked it up, it seems like a habit honed in a certain kind of affably-contentious intellectual environment (again, I’m just guessing at the causes here)

kandel
0 replies
2h48m

On the other hand, a passive student will let you go through an entire proof only to raise his hand when you finish and ask something that implies he did not understand the proof from the start.

Why didn't you say so? You wasted both of our time... I even turned around a few times and asked if everyone was following!

I know it's hard to understand when you don't understand. But I don't know how to deal with this problem and the senior lecturers don't seem to know how to, either...

t. just started lecturing

tripdout
4 replies
23h32m

I'm in 3rd year CS and I almost never go to office hours for my CS courses, because a lot of the time I don't have something specific to ask.

I go to lecture, don't understand the material, and then review my notes & the slides at home, working through the examples until I understand them.

If I get stuck on something, I typically just keep trying until I understand it, instead of stopping, remembering what I don't understand and asking it in office hours 2-3 days later, and then potentially getting stuck on something else later on in the proof/example.

Typically what I lack is is a more general understanding of the entire proof/example/algorithm instead of small individual details, and I really need to sit down with it and go through it instead of another quick re-explanation by a TA where I have to say "yeah, makes sense" before I've had enough time to think about it.

ChainOfFools
1 replies
22h38m

I recall that any problems that I had understanding _how_ to solve something, arose from not knowing _why_ it was being done in the first place. And as a result I had no entry point into the solution process.

Once inside the process of finding a solution things tended to fall into place. But looking at a problem set and not being able to guess what the first step is, thats a sign that I have not developed a means to find my bearing or orientation in the overall problem space.

what distinguishes the forgettable instructors that I have had, from the invaluable and irreplaceable ones? The ability to communicate why we are doing what we are doing as effectively if not more so, than how we go about doing it.

This is the gift that someone like grant Sanderson has, so incredibly rare and strangely opaque to (or even dismissed by) the typical college level instructor of STEM material.

eltonlin
0 replies
12h14m

The ability to communicate WHY we are doing what we are doing I really resonate with what you said. The Why and the contextual motivation is what I'm always so starved for in classes that I don't understand.
godelski
0 replies
22h7m

because a lot of the time I don't have something specific to ask.

That's okay. Most TAs/professors will still help. Unfortunately some are assholes and if you get those, sorry. But the best way to deal with this is by asking them how to get started with a problem. Or to also ask about problems you've already solved. Here's some templated questions, modify for your specifics

  I've been struggling solving this problem, can you help me get started? I tried {x,y,z} but that doesn't seem to work because {a,b,c}. 

  I solved this problem, but it took me longer than I think it should have. Can we walk through it together so I can see how you would approach it? 

  I'm having a hard time starting problems, not even knowing where to begin. Do you have any advice? 

  Can we do more problems in class?
I came from physics and am finishing up my PhD in CS, so the last one might not be as good except for stuff like algorithms. But one key piece of advice for the getting started problem is to do more problems.

In physics, the first thing I'd always start with is a force diagram and writing down my knowns and unknowns. This way I could do a form of dimensional analysis and then I had to just figure out how to convert. I then took these skills to CS and found a similar avenue. Your professors saying to write pseudo code aren't just trying to get you to do more work. This is your map.

I take this same approach when I program. I write mostly in python so this works real well. I use vim and so will have one pane open to my file and another to an ipython terminal. I can test my logic in the terminal and write code side by side. So I try to write something quick first (think "write drunk, edit sober"). Then I start writing my function and I start by writing a small description and the arguments. As I'm writing this, I find my mind often finds edge cases that I can account for. A few minutes of back and forth, and I got a pretty good function. Essentially, I am using documentation as a rubber ducky or the Feynman Method[0].

I also highly suggest working with other students. There's a lot of good reasons for this. They'll see things in a different way than you, and one of the most important tools you can build is getting these different viewpoints/perspectives. Another is you're growing your network. These are very important for jobs. And probably most importantly, you need friends and a social life. If you're kicking yourself when you're down too much, you're gonna be slower at solving problems. Sometimes the best thing to do is get up, take a walk, and come back (that's the hard part).

Also, ask questions in class. Don't worry about what other students think. You're paying for the professor. In today's age you can probably find better lectures online. So get your money's worth and ask in class. Most professors actually will be happy because when lecturing we hate having just blank stares and silence. We all dread the "does this make sense?" question and the absolutely unconvincing murmur of "yeah". Make your time count. I know it can be hard to be engaged, but also asking questions will help with that. And absolutely do not be afraid to interrupt or ask them to slow down. You're paying way to much to just sit in an in-person YouTube lecture.

[0] https://fs.blog/feynman-technique/

VyseofArcadia
0 replies
23h2m

If I get stuck on something, I typically just keep trying until I understand it, instead of stopping, remembering what I don't understand and asking it in office hours 2-3 days later, and then potentially getting stuck on something else later on in the proof/example.

Good! This is what you should be doing! Understanding comes from within, and teachers can only guide you to it. You are your own best teacher, and you will learn the material on a deeper level if you figure it out yourself. Sometimes you do need a little guidance, though, and that's ok.

Typically what I lack is is a more general understanding of the entire proof/example/algorithm instead of small individual details, and I really need to sit down with it and go through it instead of another quick re-explanation by a TA where I have to say "yeah, makes sense" before I've had enough time to think about it.

First of all, get face time with the actual professor if you can, not the TA. TAs can be great, but they're likely someone with limited pedagogical experience.

Second, you need to learn to ask questions to get to the heart of the matter. Don't walk in saying, "I don't understand this proof/example/algorithm". That gives them nothing to work with. All they can offer you in return is a quick explanation. Tell them what you just told us. "I feel like I've got a good handle on all the small details and steps in this proof/example/algorithm, but I'm having trouble understanding it as a whole." Maybe try, "can I see some more examples?" or "can I see a motivating example?" or even "what historical context led to the creation of this algorithm?"

Context can be everything. It can be difficult to understand material when it's presented to you in its modern form, and you don't see how something so logical and perfect crystalized out of the aether. If you can see a couple of steps in its historical derivation it can make a lot more sense.

Not saying this is the case for you, but I have also seen a lot of students who think they have a good handle on the small details, but it turns out they only understand gist of the details instead of the details of the details. Hell, I do that myself. If you go through every little detail and ask yourself "why?" or "where did that come from?" or "what could have motivated this?" you might discover you have more gaps in your knowledge than you realize.

Best of luck with the rest of your program! Don't be too hard on your teachers. As I explained in another comment, I spent years really trying to improve as a teacher, but it is an extremely unrewarding experience for a variety of reasons. These people are not getting rewarded for any extra effort they put into helping students, and if they're doing it anyway they deserve kudos.

gfodor
4 replies
1d2h

When I went to school office hours was where you went if you wanted to get hints on what you needed to know for the exam, not to learn stuff. It was a means to an end. Honestly I never even considered the idea that office hours could help me better understand the material. Goes to show you what the mind of a college student is like.

petsfed
0 replies
1d

I think this speaks to something in the OP that bothered me:

So much of the student-hostile elements of teaching come down to manufacturing efficient metrics for learning. They don't guard past solutions because they don't want you to study from them, but because they need to be able to report whether or not you learned the material, and writing an appropriate test question, especially at higher levels, is hard.

Likewise, you didn't view office hours as a tool to learn the material, it was a tool to game the metric they force on you to determine if you're learning in the first place.

Its very easy for the execution of modern education theory to slip into an ouroboros of perverse incentives.

godelski
0 replies
22h0m

It's a difficult balance tbh. You have to struggle to learn so teachers can't just work through the homework problems for you. So they want to give you hints to get you unstuck but not give the answer away. What would be better is to work through a non-homework problem. If this is happening to you, try to find a similar problem in the textbook and work through that one together. They'll still want you to struggle but be more relaxed with giving you hints because they aren't just handing over the solution.

deuplonicus
0 replies
1d2h

Not all college students are the same, I used office hours to get help building things, designing things, and to better understand the materials, especially when I got into silicon fabrication and electromagnetics, it was fascinating, learning exactly, in detail, how a PC works at the atomic level.

IG_Semmelweiss
0 replies
17h24m

Its also cultural. Office hours never even crossed.my.mind.

Who am i to take a full hour of the prof? Am i a remedial student? Ive never had to consult with anyone for my grades. What is this tutor everyone speaks of? My parents never had to pay for any tutor.

For me, the whole office hours its like eating raw meat. You know people do it. But you are not running on those circles, you donr go to those restaurants, and you dont even know the food name and you couldn't tell if it was raw meat even if someone threw it at your face.

Its a resource for those willing to take advantage of the system as it is set up. And in the odd case everyone does, well, eventually the service degrades enough so that one of the students eventually creates an app for it.

anthomtb
3 replies
1d

very few people would come. And then I'd get feedback at the end of the semester that I "was never available to help."

My guess is that when they "needed help" was 2 hours before the exam, upon realizing that 10 hours of cramming was not going to make up for 12 weeks of slacking off (source: I was one of these types). IOW, their deficiency not yours.

xattt
2 replies
1d

My profs in my social science-like undergrad also held office hours.

However, I had no idea “how” to ask for help or about what. In my perspective, most of the stuff could be learned or memorized if you sat your butt down.

ghaff
1 replies
23h47m

Or they totally lack foundational skills.

I tutored a group in an MBA program who were really struggling especially in more quantitative core courses. Which was not at all uncommon but this was extreme.

And it was pretty much a case of: Teach me all the basic high school math I never learned.

VyseofArcadia
0 replies
23h22m

For me, day one used to be handing out the syllabus and getting right into whatever the first topic of the course was. I taught a lot of calculus 1, so usually I'd open with a motivating example for the concept of a limit.

After a couple of years, I decided to instead spend day one on the most common egregious gaps in knowledge, so day one was how to add, subtract, multiply, and divide fractions, and then a little practice solving simple rational equations of a single variable.

godelski
2 replies
22h25m

I had the exact opposite problem when I was both a TA and later a professor. I would beg and plead for my students to come to office hours. ... And then I'd get feedback at the end of the semester that I "was never available to help."

My first year of grad school I had this problem and it made me immediately become disillusioned because the reality was so starkly different from what students wrote.

I worked in my office and would be there from 9am/10am till 10pm most days. I had an open door policy along with my scheduled office hours and 2 lab sections. I also frequently responded to emails at or after midnight. I wrote custom mini lectures for the lab sessions based on asking students (in those lab sessions) what were the things they were struggling on (and using what I ended up helping the most on + feedback from the good students that were always the "help sinks" where others would go to them for help). My own office mates complained at the end of the term (but not before. To be fair, I was usually the only one in the office (of 3) and the other two only started showing up more at the end of the term). YET I was student ranked as 2 standard deviations below the median for availability and one below my department.

I literally could not be more available! What's worse, is the department sent an email to me about how I should be more available. -__- What I learned is that these things don't mean anything about my performance. They were low ranked because the course was very hard. This is also a course I caught 2 junior CS students cheating and told the professor and I that they didn't know "GitHub" was on the internet (syllabus had a weird policy about not using the internet, but intended to basically mean don't copy GitHub or SO solutions). I had a hard time trying to not fall over laughing and I saw the prof's brain just break.

(Side note on that last part. We reported it to the department. They didn't care. It is because it makes the department look bad because the uni measures department rates of cheating by how many people are formally disciplined for cheating and so if you don't report it, you don't look like your department has a problem. We gave them a 0 on the assignment and said they'd fail if they did it again. They did anyways)

(I also had two students write that I was the best TA they've ever had and two students write that I was the worst. I learned a lot, that's for sure)

eltonlin
1 replies
12h27m

Hey sorry to hear your experience. First to clarify - it varies - the classes (usually required CS classes), usually had overcrowded Office Hours, but there were many classes in math & physics where OH was very available (and I did very well in as a direct consequence, though I was a CS major).

If students complain about availability, but never show up despite everything you did, then that's their problem. What people say and do is, unfortunately, not the same.

And I'm sorry to hear about the strange incentives for not being able to report to the department about cheating.

Initially when I started the app, I was very anti-teacher. But as I learnt more and talked to more teachers, I realized, man teachers also have their fair share of indignations from the system.

godelski
0 replies
9h21m

realized, man teachers also have their fair share of indignations from the system.

Oh for sure. I think many would be surprised how political (not in like government terms) a department can be. It can vary dramatically within a school too. My partner's department is very functional while mine seems to be continually shooting themselves in the foot trying to dislodge the previous bullet hole. But when I started, it was very functional. I can certainly say that very few professors are happy with the system and I think it is failing everyone involved. This is why in my main comment I suggested that there is real opportunities to disrupt education in a very traditional sense. Because the truth of the matter is that we started treating academia like a business and the results were probably what anyone could have predicted: metric hacking galore. It's a big part of why students themselves cheat in the first place. Then try being a grad student where you are doing half of your advisor's job, leading other students and writing grants, while trying to also do course work (first few years), TA, publish, and research. It's a lot. Everyone is overworked.

There are plenty of great advisors and departments. But unfortunately these tend to be the exception instead of the norm and appear to becoming rarer. Or maybe I've just gotten more experience. Personally, my PhD has left me with a very bad taste for academia, and I know a lot of others at other schools who have the same feelings. What's also surprising to many outside of academia is that my dream is to spend my days researching. But they cannot understand how academia nor industry provides me a environment to actually do this (fwiw, academics usually to understand).

willmadden
0 replies
1d

It's the student loan industry and lowering standards for college admissions. Most of them should not be in college.

seanmcdirmid
0 replies
1h46m

I never understood what office hours were for. In college, I felt like it was better to just spend a few hours working on homework problems, how could anyone explain this stuff to me if I couldn't learn it on my own?

I TA'd some classes also (as an undergrad and grad student) and tried to help people with their homework, but in the back of my head I couldn't help but feel that it wasn't the the most effective approach.

I can imagine some things that I don't grock yet (like some computer graphics algorithms that I just cargo cult, or machine learning in general), maybe this is blocking me from learning them, I just need to ask someone who knows it to explain the intuition to me?

jimhefferon
0 replies
23h7m

I just moved to a new college. At my prior place, I might see one student in hours in an entire semester. Here, in hours there are usually two or three each session. So yes, it varies.

bee_rider
0 replies
1d2h

Also the general level of confusion of students coming in to office hours can be a valuable part of the feedback loop.

atoav
0 replies
19h59m

As an academic educator myself I must sadly concur. Students always demand a lot of things, and then they don't use them.

That doesn't mean offering them is a bad thing, because there are people who do, but most students I know "don't have the time" and I know for a fact that they don't do shit. And I know that for a fact because I studied at my university a few years prior. Back then I was a freelancer while studying, ao when I said I don't have timw that meant I had a job. Ut for most of my colleagues that meant they were occupied with loafing around.

Not that everything and everybody needs to be efficient, but if you loaf around blaming a lack of time is a fucking joke. Maybe they are lacking energy, direction, discipline, willpower, ambition, etc. and sure, some — e.g. young mothers — are truly lacking time. But most students are not.

andai
9 replies
1d8h

Re: incentives being weird in higher education: we once overheard the director of the CS department respond to complaints about the program not being a good fit for industry by saying, "our job is to create researchers."

I recall something like 7% of students going on into research, meaning that they weren't even trying to meet the needs of 93% of their students.

touchngthevodka
1 replies
1d8h

Since when did we start making the assumption that university is supposed to be vocational?

AnimalMuppet
0 replies
1d5h

When they started teaching mechanical engineering, not just physics. And chemical engineering, not just chemistry.

They need to start teaching software engineering, not just computer science. And 93% of their CS students need to switch.

sgu999
0 replies
1d8h

What would be the percentage of researchers left if they weren't trying this hard to create them?

With a good theoretical background in CS, most of what you need for the industry can be learned more easily throughout your career.

That universities are not dedicated to teaching how to be an immediately productive economical actor makes sense. That's of course assuming you don't have to take on 300k of debt to get a degree...

humansareok1
0 replies
1d3h

Then go study Software Engineering not Computer SCIENCE lol. Idk why this mind virus continues to eat people's brains. College isn't and never was meant to be a job training program. Some majors do more then others and if that's what you want select those majors it's pretty simple.

dekleinewolf
0 replies
1d7h

It's way easier to make a good programmer out of a researcher than a researcher out of a programmer.

I have always had programming jobs (job title: software engineer) that required the 'researcher' mindset. We didn't mind teaching them our stack or even good coding practice, but if they didn't have the 'explorative/innovative mind' of a 'researcher', we could never train them up to the level they needed to be.

computerfriend
0 replies
1d7h

What do you think the S in CS stands for?

cess11
0 replies
1d8h

Between the two sentences "good fit for industry" gets rephrased into "needs of 93% of their students".

I think it would have been clearer to just stick with "good fit for industry".

Muller20
0 replies
1d8h

You don't need a university degree if you just want to learn the last javascript frontend framework, a good coding bootcamp can teach you that. Inflation of credentials is not a problem that universities created, it is an industry issue.

Gooblebrai
0 replies
1d7h

The issue is having made university the institution one has to go through to access employment.

University was meant to be the destination of future scholars.

newtone
7 replies
21h14m

it’s very hard to catch-up: once you fall behind, you have no way to review past material efficiently enough to compensate the difference - like credit card debt

It is surprising how many people working in education do not understand this. We have clear understanding on technical debt or cycles of poverty but absolutely no understanding of how learning debt evolves among students. It is so evident that couple of months into any academic session you can already see class performance on any test to be pretty similar to what it is going to be at the end of the year. Even if you know 9 months ago in an academic session that 90% of the students are lagging, you can't/don't do much about it. And the fun part is that all the debt is assumed to vanish when you start a new session. Nobody acknowledges that the kid has been collecting debt for a decade now.

Some friend from college and I developed Filo(askfilo.com), mainly for K12. We connect students to live tutors in realtime into an online class. It doesn't sound very sexy in the AI/LLM world by we are pretty proud of the range of students we have been able to help. Still lot of work to do on it though.

Xakari
2 replies
10h24m

Absolutely, there is no meaningful discourse and action aimed at leveling the playing field for students who have accumulated learning debt over the years.

At your org, These are all human tutors on the other end!? or AI ones? I'm still hesitant about the effectiveness of AI-driven tutoring.

newtone
1 replies
10h12m

We use human tutors only. Besides the academic accuracy and hallucination, I think the biggest issue is lack of proper 'uptake' currently with AI models, to continue the conversation.

The problem is, 'majority of the students do not know what to ask and how to ask a question'.

Our tutoring is AI-driven in the sense that the entire engine from matching to pedagogy-detection in a session relies on some AI model.

thinkingemote
0 replies
10h4m

We should teach people how to learn to learn. Learning to learn.

How to ask questions. This leads to how to form search queries and now how to write prompts.

eltonlin
1 replies
13h8m

"All the debt is assumed to vanish when you start a new session. Nobody acknowledges that the kid has been collecting debt for a decade now"

I think you articulate it really well. It speaks to me at a personal level because my stunted academic growth in college still affects me today, and it will never go away.

Good luck with Filo!

newtone
0 replies
10h21m

I feel you more than ever. We started identifying this long-term debt as `non-linear learning gaps'. As other folks have called out, large chunk of students in higher grades or higher ed need to understand the most basic of stuff referring back to multiple years of grades in their past. The good part is you can do that in a human conversation, go back or ahead of your current grade-level seamlessly in a conversation as your mind creates a mental-model of what you are learning. The 'going-ahead' part is specially important as it build perspective into 'why' of learning a concept.

In short, we believe the long-term debt is fixable.

ragecoding
0 replies
9h26m

Your observations on learning debt resonate strongly with many educators. It's concerning how patterns of performance tend to persist throughout the academic year, indicating deeper underlying issues. Filo seems like a valuable resource in breaking this cycle by providing personalized support to students. Keep up the good work in addressing this critical aspect of education. I will try to use Filo myself.

palash777
0 replies
10h14m

Wow, reading this article really struck a chord with me. As someone who has experienced the overwhelming feeling of falling behind in school and struggling to catch up, I understand the importance of addressing the learning debt you are talking about. It's disheartening to see how little attention it receives in educational circles. Most of the times, we tag the kids with this debt as stupid and move on.

andrewlgood
6 replies
1d4h

I am always curious regarding new education solutions. I firmly believe the US pre-college education systems are fundamentally broken most benefit from edtech (huge fan of Khan Academy).

As an MIT alum myself ('89), I am surprised to hear your description of your experience at MIT. Did you experience this across all classes and departments or only for a few (e.g. 18.01 which is required for most freshman)? While certain professors were not the most helpful, I found most of the teaching staff to be incredibly helpful when I made the effort to engage them. No doubt the focus was on attendance in lectures.

In addition to lectures there were: - Recitations - each class of the larger classes had recitations with a TA to allow smaller groups the opportunity to ask specific questions. - Office hours with the lecturer. I did not avail myself of them as I considered them useful for bigger questions (should I be a physics major if I don't understand this) rather then review of homework questions. - Bibles. Each living group (I was in a fraternity but dorms had them as well) had a library of class notes, exams, quizzes, etc for each class assembled by people who previously took the class and did well. Often the Bible author was still available for questions. - Friends/Fellow Students. MIT has an incredibly open policy on working together. My recollection is some classes could not be reasonably completed by people working alone (e.g. Aero/Astro's Unified Engineering). A fellow student may simply have looked at a topic from a different perspective which drove their comprehension.

Today, MIT offers OpenCourseWare (ocw.mit.edu) which hosts complete courses online (video, lecture notes, syllabi, additional readings, quizzes). This is an incredible resource that I have found helps my lifetime learning goals.

Again, I am always excited for edtech. I think its development needs to be tempered with an understanding of why current systems are not working. What problem are you solving? Is your solution simply remote tutoring? If so, how is it different that a job board with zoom?

nonameiguess
2 replies
1d2h

I didn't attend MIT, but the existence of OCW is exactly what made me wince a bit reading this. This guy really experienced professors refusing to record lectures to encourage attendance? That doesn't track with MIT being a pioneer among US universities in recording lectures and making them freely available, not just for their own students, but for anyone anywhere with an Internet connection. I used OCW extensively when going back to school for grad school, even though it was not at MIT. Those resources are terrific and invaluable.

butwho
0 replies
1d2h

I'm not in the USA but my universities in the 90s used to offer at least 5 online 5-year courses (CE, EE, Chemical, Mechanics and Energy) and you would have to go only one month a year for the labs. Now they offer ZERO. It wouldn't surprise me that some professors feel "offended" that people choose to follow the lectures online instead of taking the "advantage" of his prestigious person.

DabbyDabberson
0 replies
1d

('18 alum) Yeah I agree. We even had internal-only versions of OCW with more recent recordings and more material.

Although I agree that professors were not accessible, and TAs were often not that helpful. But, I never felt like I didn't have the resources I needed. I just lacked enough time.

I did feel like i would have been able to _learn_ better from a smaller school with fewer students, and teachers hired for teaching instead of research. But I still don't think that would have outweighed the benefit of the MIT community and resources.

eltonlin
1 replies
12h50m

I couldn't convey the technicality across, but the bad experiences I had were mostly localized in course 6 classes - Intro to Algorithms, Design & Analysis of Algorithms, Intro to Machine Learning. I actually did very well in math & physics because Office Hours were much more available, and they tried teaching in different formats (e.g. TEAL).

incredibly open policy on working together Yes this was mostly the case - except for course 6. Also, I - sadly - didn't have study groups. For a combination of reasons, including my flaws, I just wasn't good at making friends in college.

As for how it's different from a job board with zoom - the key is to solve the "scarcity" issue, in my opinion. Teaching 1-on-1 for n students scales O(n). If there's a way to teach n students O(log(n)) i.e. address all overlapping needs with videos that can be replayed infinitely, then address remaining personalization problems, that's the approach I believe in

hnfong
0 replies
11h19m

With the CS and ML hype in full force these days (people with CS majors increasing at double digit % points every year), it does seem like the courses you mentioned might have been taking in a bit too many students and they genuinely have a staffing problem due to sudden demand increases in the past few years..

ginbazinga
0 replies
14h19m

I'm also enthusiastic about new education solutions, and I found traditional MOOCs are 1. too long for one sit; 2. passive; 3. lack personalization. So I'm building a new solution that aims at solving these problems: https://afaik.io/, and you can learn college physics and calculus for free.

The goal is to map out the prerequisite structures among all subjects as knowledge is actually a continuous map, and let everyone have their own personalized learning path. It's visualized like this: https://afaik.io/nebula?category=brickset&id=VLlOnZLl&mode=d.... The learning unit (called brick, like Lego brick) is designed to be 10 minutes, questions are asked at the end to check understanding.

Don't get me wrong; I love OCW and I think it's still one of the best free resources online. But I do think great resources combined with AI can bring such innovative edtech solutions that everyone can have their own private tutor almost for free, and most importantly, without worrying about hallucination effects.

gnobbler
4 replies
1d5h

Are you trying to create a tutoring version of fiverr/upwork? If so, the economics doesn't seem to work.

I think you are saying that you'll create a product, charge $10/wk to customers and, through volume, give the tutor approx. $100/hr. Fair enough, but I think you're overestimating how well tutoring scales - which is not well at all.

If I'm a tutor, and I want to break even with my in-person rate, I need to satisfy the needs of 10x the people. That's A LOT of psets. If they have the same problems, i.e. they're in the same class, then they won't need separate accounts because they'll share screens. If they don't care about watching others, they'll watch YouTube. If you allow a large class to interrupt to ask questions, this will have to be managed carefully so the session doesn't grind to a halt after 2 questions.

I strongly suspect people will unsub after 1 session because they paid $10 to watch another guy get tutoring. Then again, I'm from the US, so maybe that's just here.

ebuck
1 replies
1d3h

Qualifications: I tutored math for about 10 years, I used to guarantee an A if you signed up with me, as long as it was still mathematically possible.

I think the roles of teacher and tutor are being confused in the "scale up" approach. Teachers have presentations that allow a large number of people to cover the material, with specific goals, verified by observing the class through testing.

A tutor covers the same ground, but in a much different manner. The tutor has an audience of one, and the goal is to fix errors in the student's understanding.

Only in a student that is ignoring the teacher will the tutor be the teacher. For every other student, the tutor assesses what the student's strengths are and what the student's weaknesses are, even if the student misreports their weaknesses.

Then the tutoring session provides explanation, guidance, and drills to fix the weaknesses. Often a weakness in solving one problem exposes underlying weaknesses, when that happens the session shifts till the underlying weakness is addressed, after which the session resumes on the upper level problem where it paused.

The rest of the tutoring is giving the student ample work which is completed under supervision, until the student builds skill. Good problem generation is often overlooked. Good problems rarely are the same problem with different numbers, good problems challenge the student to use the tools in a variety of different scenarios and problem formats. Eventually the student will tell you they understand it, and their work will support their claims.

So, watching someone else being tutored is like taking a bespoke suit and putting it on someone else. Yes, it might be wearable under some circumstances, but it's not going to properly fit the person the suit (or tutoring) was tailored to fit.

dgacmu
0 replies
22h3m

I'm a professor, and I wish more people would read your comment - and then take more advantage of tutoring, be it peer tutoring, office hours, or other resources. Tutoring is awesome and very time-efficient for the student. But also very resource-intensive for the same reason.

eltonlin
0 replies
8h11m

Maybe you're right, but I'd challenge your middle paragraph.

When there are 10 students, they often have some common misundersetandings that can be addressed together e.g. they all want a review of Taylor's theorem needed to do electromagnetism, and they forgot it from high school. So I argue there is a lot of room for reusability of explanations.

"pay $10 to watch another guy get tutoring" The format async., so if person A asks a question, person B doesn't have to sit and watch. The same way it works on Stackoverflow - if someone asks a useful question that you wanted to ask, you can reap the rewards of the existing answer.
andrewlgood
0 replies
1d4h

I agree with your assessment re this being similar to fiverr/upwork. I do think the economics can work for tutors that generate high credibility/success ratings. Codementor.io focuses on providing programming mentors to help people solve a very specifc question or class of questions. It is not meant to be a weekly scheduled session, but rather an on-demand service.

Additionally, many of these services could be useful as incremental money for the tutors rather than the primary income source. Definitely would pay better than UberEats!

julianh65
3 replies
1d2h

I'm going to give this a go, I've been self studying linear algebra, and there are certain questions that I can't seem to find good answers or explanations to online that ChatGPT or Claude cannot answer sufficiently. The one concern I have with this, is the async nature of it. One of the reasons I like using ChatGPT or Claude (or a real human if I can find one) to learn is because I can ask clarifying questions and try to relate what they're saying in real time to another concept I'm familiar with. Curious to try it out but I think that an important part of making the videos good is that the student / learning needs to provide a good prompt so that the response can be customized to directly address their area of confusion.

eltonlin
1 replies
14h51m

Ah I see why you'd want realtime interaction. In earlier versions there used to be voice chat, so when the learner-and-teacher happen to both be free, you'd talk in real-time. It was removed to focus on the async video aspect, but hearing what you said puts that back in my radar...

When you mention "good prompts", is that in the context of AI, or with asking questions in general? Is it easy/hard to create good prompts from your past experience?

As for Linear Algebra, I noticed you didn't sign up - did something change your mind? Feel free to be brutally honest as it helps me make real progress! If it's easier, feel free to call +1 503 250 3868 (Tokyo time zone), or email eltonlin@alum.mit.edu anytime!

julianh65
0 replies
1h50m

Appreciate you following up! I plan to sign up over the weekend which is normally when I do my longer blocks of studying and have more questions. I think that what I mean is that sometimes if I ask for an explanation about a topic or question, I don't get the magical answer that answers my question after my initial prompt. That is to say, I typically need to ask 5-6 clarifying questions to an LLM to get the explanation that makes it click for me. This can be by asking for more examples, asking it to clarify certain points, fitting an example to another topic I know, etc... I think that while I was talking about it in the context of LLM tutors, the same logic also applies for human tutors as well. It's very hard for me to precisely formulate a question in one-shot that will get the right answer as you don't know what you don't know. And the reason I'm asking LLMs or a human in general is because I can't find a good explanation on the internet as it's typically something very niche or precise. Which is why I think it's good to have a back and forth.

cosinetau
0 replies
1d1h

Hi - math student here too. Have you asked any image AIs for graphical representations of these Linear Algebra problems? How did those attempts go?

Asking because visualizing the problem scenarios was something that did not come right away for me, and I had to spend a lot of time in the tutoring center to build it. I believe the right prompt might yield the right thing.

throwaway81523
2 replies
1d7h

Is this a video version of stack exchange, where you pay to get your questions answered on video?

Anyway I've given and received actual tutoring and imho it just can't be done on video. E.g. in a real session, the tutor might give you a problem to solve (pen and paper) while they sit next to you. You start working on it and when you hesitate somewhere, or if your approach to the solution starts going wrong, the tutor immediately notices what is happening and gives you some guidance.

I suppose this video product could be of some use to some people but I can't imagine it being like tutoring.

abdullahkhalids
1 replies
1d

Yes, the teaching process is often a two-way and long conversation.

Stack Exchange works in the case where the questioner already has enough mastery over the material to ask a well-posed question. But many students don't even have the mastery and they never get around to asking questions on SE, and if they do, that question is quickly closed. So you never see those students online at all. Maybe on more niche forums where there is more back and forth.

As a former physics prof, the students most in need of help are precisely the ones who can't ask questions. In that case, someone has to sit with them and with a back-and-forth figure out what they don't now. Then teach them that.

So I agree with you. This is a product mainly for above average students trying to understand a bit more. No problem with that. You can choose your customer. But I want to highlight that any sort of teaching methodology (edtech or meattech) will serve some students at the expense of others.

eltonlin
0 replies
13h0m

Yes you guys are right. This "video-based Stack Exchange format" will have clear disadvantages. You can't observe the non-verbal queues in-person for example that helps you discover the root of their misunderstandings etc.

Any sort of teaching methodology will serve some students at the expense of others Yes, I admit this isn't helpful to all types of students, though I wish it can at least be helpful to an initial set of students
promiseofbeans
2 replies
1d7h

Just a note, looks like you're using google fonts (or similar) for your icons, so people (i.e. me) with strict unlock origin settings will just see big text labels. Here's a screenshot https://i.ibb.co/qFc0X2g/image.png

leosanchez
0 replies
1d5h

ublock right ?

eltonlin
0 replies
13h13m

Ah, which alternative icons would you recommend that wouldn't get blocked by strick unlock origin settings?

pbhjpbhj
2 replies
1d8h

Couple of UX focussed comments:

The site was impossible to use on FF mobile without using desktop-view. https://www.explanations.app/dat153qkJtTcHf9Fbm1B/overview, for example has the content frame as a slither on the right.

Signing in with Google and getting a different name is disconcerting.

It's not clear what a "server" is?

I searched "electromagnetic" and "electro-magnetism" but got no results ... "Electricity and Magnetism" is one of the videos though. Robust search is essential.

On one page there were a few "untitled" blackboards; allowing them to be untitled seems bad. They were empty, suggesting some moderation/UGC vetting will be needed.

It's important for college-students to be able to determine the college-level content from the other content. You say it's an open platform, how does one readily see only college-level content?

eltonlin
1 replies
12h36m

Thank you for the thoughtful comments, I appreciate it.

FF mobile: you're correct, this is not mobile responsive yet - I will work on a future version to fix this.

Different names with Google Login Is it that it shows "feynman-mvp" which is different from explanations.app, or is it that your name is shown to be different?

What's a "server" I was trying to re-use the same concept from Discord "servers", a community of people. Which word would you use instead?

Robust search Sorry that you got nor esults - yes the search is pretty terrible. Untitled/empty blackboards are not great too - moderation and vetting will probably be needed

How does one readily see only college-level content In this current version, all the math & physics "servers" are college level (so manually vetted), and everything else is categorized in the "everything else category". Is there a better way to do this?

Thanks again!

madacol
0 replies
2h10m

It's very confusing how you use ">"

It is generally used to quote literally a passage from the person your replying, for example you wrote this:

FF mobile: you're correct, this is not mobile responsive yet

and here, in a new line, people usually respond to what is quoted above ↑

passwordoops
2 replies
1d8h

Anything to democratize education is a good thing, so I'm really happy to see this and hope it takes off!

spurgu
1 replies
1d7h

Wouldn't "decentralize" or "diversify" be a better term than "democratize" here? I don't really see the democracy involved as opposed to simply having a broader disaspora of education.

Agree with your point though.

wccrawford
0 replies
1d6h

When people say "democratize X" they mean to bring it to the masses, so that it's accessible to everyone.

I wouldn't say this decentralizes or diversifies anything. This is 1 website, making it a central point. And it's just the same math. Diversity, in any form I can think of, has nothing to do with it.

SimonDuerr
2 replies
1d5h

Interesting concept, and a unique landing page! The 3D carousel confused me a bit, especially when it kept rotating while I tried watching one of the videos.

eltonlin
0 replies
13h20m

Ah OK, so I better make the carousel pause when a video is playing then!

brianbreslin
0 replies
1d2h

this threw me off as well. I thought hitting play would pause the rotation.

voiper1
1 replies
1d8h

That's a compelling story, I wish you the best of luck!

EDIT: Explore AI as part of this (e.g. using RAG), many have found it super useful as a tutor for personalized explanations when you don't understand the textbook. If you have it grounded in the textbook or other sources, it should have fairly minimal hallucinations.

pbhjpbhj
0 replies
1d8h

Given the absolute errors I've found in LLM responses, like opposite definitions for opposing terminology, I very strongly counsel against using LLM output for learning.

I like Phind, because it gives links with human-authored source material. I use it to test my knowledge and when it doesn't agree, I seek reputable sources.

People seeking extra tutor support can't themselves tell, in all probability, when there are subtle errors in the LLM outputs and these could be really destructive to proper understanding.

vikR0001
1 replies
1d1h

This is very cool. Note: the chalk-on-chalkboard sound comes to my speakers as harsh static.

eltonlin
0 replies
12h3m

Ah I'm sorry for that. That was a bug from an earlier version that saturated the javascript event loop and caused the microphone to not work properly. This should be fixed as of last month's version.

Sorry again!

v_london
1 replies
1d7h

Hey Eltonlin, this sounds really cool and I hope you can make an impact in improving access to education. Some time ago I wrote a blog post titled "Professors as Creators" which discusses the impact the Creator Economy could have on education - how in other fields (e.g. writing and videos) creators are increasingly leaving large organisations like newspapers and film studios to create content on their own using platforms such as Substack and YouTube, and how this could impact education next. I think you might like it as your platform designed around these ideas.

https://medium.com/age-of-awareness/professors-as-creators-h...

eltonlin
0 replies
15h14m

I just read it, I agree with your points wholeheartedly.

tines
1 replies
1d3h

This is exactly what I've always wanted, hope it grows!

eltonlin
0 replies
13h33m

Thank you! If you want help setting up you can always call +1 503 250 3868 (Tokyo time) or email me eltonlin@alum.mit.edu :^)

tayo42
1 replies
1d5h

but very expensive e.g. $100/hour

Yeah getting specialized human help that is worth while for anything is expensive. I tried to being a tutor/mentor on some site, forgot off the top of my head. I'm pretty sure in and out or target pays more.

100/hr seems right to charge. Probably still not even enough. If you break it down, you spend a ton of time unpaid doing prep and searching for clients, the work it self is inconsistent, you pay taxes.

It also felt like a race to bottom on prices, where I think 3rd world country people were working as tutors.

I also just got asked to answer homework questions for cheap.

butwho
0 replies
1d2h

I also just got asked to answer homework questions for cheap.

My exact experience as a tutor. The clients do not want tutoring, they want an afterschool homework program.

srameshc
1 replies
1d6h

This looks great. My only suggestion is to be more specific and clear on pricing, rather than 1/10th of the price.

eltonlin
0 replies
13h23m

Ah thanks for your feedback. I was worried that if I put $10/week, then it's not immediately obvious that it's a magnitude less expensive. Would it perhaps work if I highlight/emphasize the $10/week in the description below on the website? Or do you think it must be in the title?

rodneyzeng
1 replies
1d

Can this service expand to include coding/programming?

eltonlin
0 replies
13h25m

Which programming subject do you want in particular? If it's coding syntax heavy, then the blackboard format might not be suitable. But if it's algorithm-heavy, then blackboard visuals could be very helpful.

Feel free to email me eltonlin@alum.mit.edu or if you prefer calling https://cal.com/eltonlin1998/15min

(BTW the things I described like overcrowded Office Hours etc., restricted access to past problems, were all Comp. Sci classes - Intro to Algorithms, Design & Analysis of Algorithms.)

eltonlin
0 replies
15h16m

Ah I couldn't figure out how to linkify links - how did you do achieve this?

n9com
1 replies
1d5h

Well done on launching.

1. The price should better highlighted on the website, it's not very visible. 2. How do you even sign up to the plan? Add a pricing/purchase page. 3. $10/week is dirt cheap 4. If you're serious about this, please invest in better UX/UI

eltonlin
0 replies
12h3m

Thanks for the concrete feedback - OK I will make the price more visible. As for pricing/purchase, it might not necessarily make sense to students - from what I understand, they want to look for the right subject, then the teacher, before deciding to subscribe and pay.

Where should I learn about better UI/UX?

grok22
1 replies
1d3h

I think, as a kind of "get an explanation for a specific thing" this is a fantastic addition to the world of teaching/tutoring and hope it succeeds.

I tutor my son in engineering level subjects some times, but I find that I need tutoring myself too to understand some of the things :-). There are various well-known homework-help sites where you can post questions and get answers from "experts", but, none that I know that use a video based approach for the response.

As an aside I think we need more videos that explain how to approach solving specific types of problems -- not just the answer, but how to tackle breaking down the question and what to look for in the question to decide what theoretical aspects apply to solving the problem. I guess once a person has sufficient theoretical knowledge, they get an intuitive understanding how to do this, but engineering coursework doesn't leave too much time to digest a subject in depth and any help provided to minimize the time to internalize concepts/insights is always good.

eltonlin
0 replies
13h14m

Thank you for your kind remarks. As for the need for more videos on the "approach" of solving problems, that really resonates with me. I always wanted to learn how those mathematicians "see" matrices for example, how they "think" of what to do in the first place, etc.

If you're ever interested to try teaching/learning on explanations.app, I'm always here!

franklin_p_dyer
1 replies
1d2h

I'm curious - do you have any plans for doing "quality control" as more tutors join your site? What will you do if someone joins as a tutor and is giving incorrect/misleading answers to students, and getting paid for it?

quantxx
0 replies
1d2h

Somethings I can immediately think of, but definitely not enough to address the issue.

1. Anonymous student ratings, if the solutions are obviously unclear or unhelpful to the student. A history feature would need to be there so the student can go back and mark a question as unhelpful or wrong even if he realises this after a week.

2. You'd need to separate unclear or unhelpful answers from plain wrong maybe? I don't know what's the line you'd need the draw but if an answer is deemed to be false, the student would be eligible for some in-app credit that he can redeem towards his next question/subscription.

3. Random cross verification of answers? Other teachers would get to judge a random answer from another teacher and provide feedback. Students would be notified automatically if an answer is revised in this way, while also providing the correct solution. Teachers can have an infraction counter or something for the same.

fn-mote
1 replies
1d3h

The blog post listed on the site is wonderful.

https://eltonlin.substack.com/p/a-story-about-802-teal

I feel like it barely connects to the rest of the story, though.

The founder basically says small group work interspersed through the learning process was a major factor in their success. I believe it.

The professor says the TEAL system worked because it forced students to regularly invest time in the class, so they realized early on they needed to do more - rather than putting it off. I believe it.

What I do NOT understand is how Explanation.app is offering either of these crucial ingredients. It seems clear this site will attract the usual desperate, last-minute-cramming crowd.

The problem of college students undervaluing their education seems to be a major obstacle to the success of this venture. MIT estimated yearly cost: $85k. Somehow $100/hour for tutoring isn't worth it, though? I agree the valuation is more complicated than I want to write here, but in the end it still does not look good to me.

eltonlin
0 replies
8h18m

You're right, the blog talks about one thing, the product does another thing.

The reason I conducted those interviews was that I wanted to understand the good intentions behind instructors who I disagreed with, especially instructors who put an extraordinary effort to change things(even when it was unpopular). That's how I started to understand the strongest arguments for mandatory attendance, and a synchronous learning experience. That's also how I learnt first-hand that the key incentives problem "what's the reward for teaching well at a research university?"

Having said that, through talking to other students, especailly the busy ones, who live off campus, have a part-time job, and via my experience, I understand first-hand the need of efficient options that don't require commuting and waiting. explanations.app is async. and remote for that reason. There are lots of existing async remote options, but they are textbased. I hope providing a visual async remote option, means that it can attract students who genuinely wants to learn things properly but also efficiently.

caddemon
1 replies
1d6h

The Piazza post isn't publicly viewable, but I'm surprised you had this experience at MIT. The majority of classes I took had a ton of materials online and there were few instances where I felt I needed to go to class purely because of an attendance policy or a reluctance to share materials. Office hours are designed to be 1 on 1 for educational reasons, but other students collaborating at office hours (or outside of them) was encouraged on psets. I wouldn't expect any of this to be worse post-COVID either. Maybe a course 8 issue? But I'd be sad to hear this was the case now with 6 or 18 because it certainly wasn't in the past (and MIT was way ahead of peer institutions on this historically, in large part due to OCW)

eltonlin
0 replies
12h10m

Ah I didn't realize the Piazza post isn't publicly viewable, thanks for telling me. I recreated it as a screenshot now https://imgur.com/gallery/GtuF78M

This is a course 6 issue mainly. Course 8 and 18 I've had generally good experiences with (but I'm a course 6 major).

As for OCW, I think it "looks" complete, but it's actually not very useful because so much material is missing e.g. no pset, no solutions, only lectures but no notes, only lecture notes but no recitations, etc. etc. which is why I know Aayush and Ashay are working on https://mitsoul.org/

Anyway, thanks for your comment!

a5seo
1 replies
1d4h

You should read Imposters in the Temple by Martin Anderson for a full critique of academia’s pedagogical failures.

eltonlin
0 replies
13h21m

This book looks absolutely riveting - just read the preview on Amazon. Which e-books platform did you buy it from?

ykonstant
0 replies
1d5h

[Self-promotion] If, nevertheless, you need personalized tutoring in undergraduate or graduate mathematics, I am providing private lessons in person and online.

From my perspective, the benefits of one-on-one tutoring is that we can talk live, select material of your interest/needs and problems to target you specific weaknesses, solve problems together (like pair coding!) and immediate feedback on progress.

If anyone is interested, here are my services:

https://drive.google.com/file/d/1GtBrEfwT34y-61SZHaDRCFB9wPE...

and my CV:

https://drive.google.com/file/d/1woxVNcJ4AmT7dD2WEnYr9BHEEY7...

techFFF
0 replies
5h22m

Good plan,I want to try it Now

peppertree
0 replies
1d2h

I would suggest getting the teachers to put up a catalogs of frequently asked questions to help with seo, and charge for custom videos.

manzanarama
0 replies
23h26m

I don't understand what "because" is referring to...

"explanations.app has visual forums where university grads will properly resolve all your questions for $10/week, because the video explanations also benefit other students."

lolbullshit
0 replies
1d2h

in nature it is the restriction that affords growth. Not growth itself. Growth is like cancer. Restriction is like selection. Upvoting alone also doesnt outline content that is toxic, incorrect and makes it harder to find content that we are interested in. Afterall, you have a title that explains if the content is of our own intrest fairly well.. the upvote is a popularity contest. Downvoteing alone allows our own allows ourselves to decide what We care to endeavor in.

The way this supposed enlightened website grays text that speaks contentously as a way to tell people, "he is being negative we think" seeks to reenforce the popularity contest sort of aggregation.

I wish It didnt happen here. This persons post got to the top but HN wants to still put their own 2 cents into the matter.

ksvarma
0 replies
1d1h

Really great work here. Looks very promising. All the best!

impendia
0 replies
1d5h

I wish you luck in your endeavors!

I am a math professor, and FYI your (1)-(3) are not common practice in math departments which I've seen. Math professors and TAs are generally happy to encourage students to learn on their own. As for administrators, if they tried to impose (1)-(3), then at most US research universities they'd get the raspberry.

Of course, MIT might be special -- it sounds like a ton of students show up to office hours, and I can understand how MIT's math department might want to manage their TAs' workload. Unfortunately, the TAs are there to do research, and they are not required or expected to go the extra mile in teaching. Wanting a long one-on-one session after each lecture... yeah, unfortunately no TA is going to see that as their job, and any TA who does will get behind on their research and burn out quickly.

You probably know this already, but MIT is highly atypical. Depending on what your target market is, I'd research what is going on at state universities. Google "student success in calculus" to get a sense of some of the ongoing discussion. If possible, find a professor or TA at such a university, and offer to take them lunch somewhere nice if they'll talk about their experience in teaching service classes.

A lot of universities want to do better in this regard, especially at the "service course" level, e.g. calculus and below. And there is lots of money out there, although university budgets are weird and the politics are byzantine, which means that your efforts would need to line up with some administrator's agenda to get a piece of it. Good luck.

iamlearningai
0 replies
8h6m

This is exactly what I did.

gumby
0 replies
22h39m

This looks quite interesting and useful!

But I wanted to comment on some bitterness in your post:

the school went the extra mile to ensure I don’t have the resources to learn on my own,

The things you list are department or even course-specific. For example when I did thermo (forget the course number) they had not just lecture/tutorial/office hours but handouts about things people found hard to understand ("mud" as in "clear as").

Also study groups are life saving. Would not have passed Unified without it.

It turns out that universities succeed by being prestigious, not by teaching well.

I can't speak for other institutions, but don't think of MIT as a school (total spending on education is only ~15% of revenue and run at a slight loss). MIT is a huge government (and industry) research lab with a small school bolted on the side. Most of the staff is research staff; most of the faculty seem to resent teaching, and undergrads in general unless they can contribute to research (this somewhat overstatement ignores some faculty who love teaching and students -- but they are a minority).

You can get a fantastic education at MIT (I sure did, as did some friends) and you can get a terrible one (I have many friends who say that's what they got). The difference has nothing to do with brains or aptitude, it's just if you happen to accidentally get into the "groove". From visiting friends at other schools and looking and my kid's education I can say MIT's experience is really different, and not necessarily better.

graycat
0 replies
21h22m

cheap alternative

Here from my experience is a still cheaper alternative: My experience? From 9th grade on, concentrated on math. In college also took a lot of physics. Got a Ph.D. in applied math. Taught math in two big universities. Applied math to US national security (mostly Navy work), FedEx, and computing.

From that some advice:

(1) Get a good outline of math, that is, a list, with short descriptions, of the main subjects, and for each subject get 2-3 of the best books. One source: Course descriptions at the Web sites of some of the best universities. In this, draw ONLY from the very best, e.g., Princeton, etc.

Fact of life: Mostly the teaching and books are, in a word, junk. The worthwhile teaching and books are under 10% of the total. It's simple: A big fraction of the profs are much less good than the best but still teach and write books. Sorry 'bout that. Soooo, did I mention, draw ONLY from the very best. For cheap, pay no tuition. Get the books used, cheap.

(2) Study the books, and work a lot of exercises. Sometimes go through quickly and get only some high points; other times go slowly and try to understand nearly everything there. Sorry, at least in math, it's "not a spectator sport". Maybe a course with lectures can help, but really to do well with the material still have to do that sitting alone with the books. Sorry 'bout that.

(3) Work to understand the psychological, social, national, economic, power, political, competitive, publicity, and business aspects -- these have important effects, and the real world of the subjects is not 100% pure knowledge. E.g., get some understanding about how US/EU academics works. Use this understanding to help make decisions about career directions.

(4) As you proceed, think at least a little about how might do some original research building on what you have learned. Eventually you will find some results to publish, and that will open some special doors for you.

godelski
0 replies
22h38m

It turns out that universities succeed by being prestigious, not by teaching well.

As someone that's gone to mid schools and has lots of friends that went to top schools, I was surprised to find that I think I got more education out of my experience. BUT they certainly got far more opportunities (internships, RSUs, ability to publish, networking, etc). I could sit down and solve a problem faster than them but they had more hands on experience and had several internships and research opportunities I envied. It was kinda odd but made sense.

But it took me far too long to realize something rather obvious. Education quality __should__ saturate and likely have a bias towards the mid schools being slightly better at teaching. The reasons? Well, this is conjecture, but I'd like to hear other's thoughts.

Now I still think in general the top schools are probably better. But I think the difference is much flatter than many assume.

- There's just far more graduates per year than academic positions. Even at the Community College I started at I had a Harvard educated math professor. At my Uni (I did physics), we had a Harvard, a few MIT, a few Berkeley and others. I guess it should make sense that it should trickle down if the market is saturated. Differentiating candidates can be a messy business sometimes and can come down to nuance if all of them are top level. MIT alone graduates about 50 PhDs in physics a year and across the US there's 200 tenure track positions. Berkeley does about the same, and so 2 schools could fill half of all positions (without getting into specialty and all the nuance like living location and spouses. We're guesstimating here).

- Possible bias towards mid schools can happen when top schools are highly focusing on research. At a R1 university, professors are more judged by their research than their teaching skills, so it becomes a secondary priority. They care more about their grad students, research grants, etc. So you get graduate students teaching (which can be very hit or miss) or often checked out professors. Of course, there are also lecturer positions and those can make this fuzzier. And we all know stellar people like Gilbert Strang. Basically if you want to do the best research you're going to aim at the top schools with the top funding who will enable you to do the most. But if you're more theoretical or care more about teaching or where you live, you won't prioritize that as much.

- Does metric hacking or metric pursuit play a role too? Academics is very cut throat and competitive. There is a big publish or perish paradigm and a major focus on work being novel. You can probably imagine it can be difficult to publish work that is highly novel and successful while doing it quickly and continually. We've seen plenty of works that get through by "novelty through obscurification" as well as flat out frauds. These are far from the majority of cases, but certainly we should recognize that the system has a pressure that encourages this behavior. Which what I'm getting at is that if you want to sit down and study a long term problem or want to research unpopular ideas and teach you probably aren't going to end up at a top university either. So question is how this factors in.

Overall I think the first point is really the most critical, but I think the others are worth considering. I'm sure some other people might have thoughts and I'm very open to hearing them. I think we can do a lot for academia by destroying this notion of prestige. At the end it matters what people can do, though unfortunately that is often hard to measure (otherwise hiring would be really easy and fast). But prestige is self-reinforcing as it makes those schools more likely to get recruitment, internship opportunities for their students, and grants.

I do really think there is a big opportunity for disruption in education. I do love remote and online education, but I still do not think it is a substitute for what can be done in person. But I think this disruption is much harder and requires a cultural shift. Assuming the premise is true. But some of that cultural shift can be things like taking interns from lower schools, because those networks need to be formed.

fastball
0 replies
1d8h

I'd recommend looking at PatrickJMT's videos[1] – he does the same as Khan Academy for math but a lot of his content is undergrad and above.

[1] https://www.youtube.com/c/patrickjmt

chrisjallen
0 replies
1d6h

great idea, definately see the gap in market, well done for pushing through :)

RomanPushkin
0 replies
1d3h

Can you write a blog post "What I wish I knew before going to MIT/college"? I find the reason you made this app very intriguing. I want my son to know thing you've explained here. If not - it's not a problem, thanks for your work! Best of luck!

PartiallyTyped
0 replies
1d2h

Recommendation; instead of "cheap alternative", you could use "more affordable" as that has a positive connotation whereas the former has a negative connotation.