The MIT Challenge -- Learning 4 Years in 12 Months (Without Taking Classes)

[MyDavidsun]

thanks for the video

[TheSanovita]

interesting video!

[MultiYolgezer]

you should focus on specific subject no cs student take all those. drop these ones 6.01: 8.02: 6.01: 5.111: 7.012: 6.002: 6.013: 6.041: 6.011: 6.450: 14.01: 14.02: 24.222: 24.241: 24.244: 14.23: 14.20: 14.48J:

[MultiYolgezer]

6.034: Artificial Intelligence 6.801: Machine Vision 6.837: Computer Graphics these courses are not high level for ai type of work you seek to excel in I would recommend you to replace some of your courses with the ones on youtube. stanford machine learning stanford linear dynamical systems stanford convex optimization nptelhrd image processing v=CVV0TvNK6pk nptelhrd neural networks nptelhrd probability continues...

[TheMITChallenge]

@MultiYolgezer - Yes--I've seen some of the Stanford lectures on Machine Learning--very good stuff. I'll definitely have to follow-up with it once the challenge is finished!

[samsmd]

I enjoy reading your posts Scott, keep it up!

[l337z0r]

Lots of negativity, especially from current MIT student. I'm getting the sense they feel threatened by this. You'll learn a lot by doing this. Perhaps not as much as an actual MIT student studying the same things over a much longer period, but nonetheless you'll learn a lot. And that's respectable.

[ScottHYoungVid]

@l337z0r - Well it's an experiment, using MIT as a benchmark. There will undoubtedly be drawbacks from pursuing my approach, but there are major advantages too (namely that I'm doing it in 25% of the time and at 1% of the financial cost). I think a lot of the MIT students' criticism raise good points about those drawbacks, which I've already been adjusting to compensate, but they hardly invalidate the entire experiment.

[cschick12]

Publicity stunt indeed. What bothers me is how you publicize this as completing courses needed for a degree at MIT in a year. I have said degree, I have taken these classes. One of your "finals" was not even a full final, but rather just a section of a full final that didn't even cover half of the topics in that class. It is totally fine for you to attempt what you are doing, to try to attain knowledge and what not. Well then publicize it as that and acknowledge all the shortcuts you are taking.

[azaznator]

You think MIT courses are like OCW?

Think again. I looked at the 7.012 final you took, ours last semester would rape the shit out of yours. Seriously, your final is easy as fuck, I could've probably done it back in high school AP Biology.

[andreeaweed]

interesting video!

[A3Marcia485]

duuude

[JgJulianna565]

nice vid

[1Pelito1]

I'd love to see your proofs on linear algebra concepts

[1Pelito1]

You could learn a lot, but to be as successful as Cal Newport on MIT, the only way to go is by having intense focus for a small number of hours through the day. In other words, deliberate practice needs more time, if you want to become world class.

[TheBastardOfOrleans]

What a slap in the face to actual MIT students.

[Paroxymic]

@TheBastardOfOrleans except he deems a D or 50% to be passing. he'll be in the bottom 10% of students in terms of subject mastery for each class. Also, 6.005 has seven week-long projects that he won't be able to grade himself because you're not supposed to write or know any specifics of the final testing suite. Oh, and that course is req. for a 6-3 major.

[vassiliki130572]

Good Luck

[kennybastani]

Good luck. Remember to get enough sleep so you can retain the information.

[Liaofulin]

Sounds stupid. You are taking tests you find online then posting the answer key so we can see how well you do? Are you serious? What's the point? If you want to impress someone, actually finish a real degree in 1 year.

[skFriberg]

Pyropuncher sent me! <3

[dreamwishlove16]

haters will always hate lol

[flwrstrm1618]

hey scott, what software do you use to increase the speed of the lectures, so you can complete them on monday and tuesday?

[daambassador1]

This seems infeasible. Classes are the least of your worries when it comes to learning to program. It takes what seems like endless hours coding, and very often getting nowhere for hour after hour, day after day. Learning to code (which you will need to to do pass your tests), on top of all the useful but superflous subjects, seems extremely unlikely, unless you're abnormally quick/smart.

[ScottHYoungVid]

@daambassador1 - I've done programming before, as I mention in the FAQ. My goal here is to learn computer science, so I'm not expecting to become a master programmer, but to have the theoretical basis to work on interesting projects. I've taught myself various programming languages in the past, so I don't really feel out to prove that.

[raefoe]

@ScottHYoungVid - No, I'm not suggesting you replicate MIT. Just saying that the classes alone won't give you any practical, applicable experience.

[ScottHYoungVid]

@raefoe - Theory underlies practice, however. Understanding machine learning, graphics, compilers and other theories give you a stronger basis to work on more interesting projects.

[raefoe]

Not to mention that the most valuable parts of MIT are probably the research opportunities and connections with profs and other students. Most people who have a head for it learn how to build software on their own.

But really, good luck. I'm not trying to discourage you, I just think you should go about this in a more direct way. Can't imagine MIT stripped down to just the exams and assignments -- that would be hell, and not for much gain.

[ScottHYoungVid]

@raefoe - True--but my challenge doesn't have to perfectly replicate the MIT experience on every dimension in order to be a success. There are definitely pro's to going to MIT, but not everyone can go to MIT or wants to, so I want to explore an alternative.

[raefoe]

I'm a senior in CS at MIT, and I can tell you that if you're skimping on the assignments or labs at all, you're not going to learn much. And that's just the bare minimum. The degree is worthless if you haven't been concurrently learning from your own projects. Those classes are just a way to rigorously learn the foundations.

This stunt doesn't sound like it'll be much fun. I'd spend the year independently learning how to code instead...that's what most of us nerds did in middle school xD

[stevepavlinadotcom]

Good luck, Scott! You can do it. :)

[ScottHYoungVid]

@stevepavlinadotcom - Thanks Steve!

[jens009]

Great attempt, but you won't be able to fully understand and apply CS without any of the labs and coding for the classes. The exams are also easier than the Problem sets they hand out weekly, so you're definitely missing out on a huge portion of the class. As for the other Text-books based and low level classes you can definitely do this. Still, good luck. =]

[ScottHYoungVid]

@jens009 - I have the problem sets too for most classes, and I practice on those too when preparing for the exam (just like any real student would). Coding is necessary certainly, I'm not sure why people assume I won't be doing any.

[jens009]

@ScottHYoungVid Alrighty. Best of Luck as always. Here's the AI page we're using right now: ai6034.mit.edu It's updated weekly but they put down the material (including the code) once the semester is over.

[DvineDragoon]

All the best, Scott! I'll be following closely =]

[SlowY00]

Why does Welljohnston go so hard on Scott Young? He explained that programming is NOT the goal of CS, which is true. Not that you can stop Scott from challenging himself, but why are you trying to discourage him? :)

[kluikens]

Right. Post another video when you write your Compiler in 11 days.

[ineedmunchies]

This will be extraordinarily difficult, but I wish you good luck!

[TheBastardOfOrleans]

There is no way you'll do this man. What a slap in the face to MIT students. You really think you're smarter than them. Fat chance bro. 

[deadapostle]

Yeah... I'll see where you are in a month.

[ispeaktroll]

So like, that's a little under 3 classes a month, which means you'd have to spend 1-2 weeks on each course... It is literally not possible to learn some of those courses in any sort of depth in that much time, even if you spend your entire day on it.

[lived666]

All in all, it would be an impressive feat to accomplish, but it's not exactly a Good Will Hunting like achievement.

[ScottHYoungVid]

@lived666 - True, but it doesn't need to be. I simply want to show people that there are alternative paths and methods in learning, not that I'm somehow a savant who can do things out of the reach of normal people.

[lived666]

CS is a compact and finite field. The prerequisite classes are universal among all majors. If you already grasp the core concepts in CS such as OO, compiler theory, machine learning, linguistics, etc, then you shouldn't have any trouble accomplishing your goal. In other words, if you've learned this once, or at least have been introduced to it, this challenge is not as daunting as you make it out to be. MIT courses are not any more magical than that of other universities'.

[kunjaan]

If this is ploy to get subscribers and publicity for your blog well done. Because any CS student will tell you that the amount of time you are planning to dedicate on subjects like Machine Vision will not be adequate. Why dont you cut down on your courses? Just focus on one aspect of CS. For example say I am going to major in Systems. This will increase the chances of success. You will be a better person after that endeavor.

[ScottHYoungVid]

@kunjaan I could do fewer courses, but that would be less fun. :)

[kunjaan]

@ScottHYoungVid No it wont. I would really like to see what you eventually build after learning such wonderful techniques in systems.

[ScottHYoungVid]

@kunjaan - True, but the goal of MIT CS isn't just to build things, it's to have a deep understanding of how to solve higher level problems, as witnessed by the high % of courses which have over 50% of the grade based on a written final exam, with actual programming assignments being relatively minor. Building things matters, but I'm here to learn CS, first and foremost.

[kunjaan]

@ScottHYoungVid You do realize that a lot of your courses are either engineering, system oriented or lab centric. The true value of those courses are the rigorous assignments or the labs. It seems that you haven't even glanced over the course at all. I would have given you the benefit of the doubt if you had proclaimed that you would do the assignments. I am sure you will realize that the interesting problems, really amazing insights are hidden in the homework questions and the labs.

[ScottHYoungVid]

@kunjaan - It depends on the course. For some, especially the higher-level CS ones, I would agree with you. For others like calculus, linear algebra, discrete math or even analysis of algorithms, they are mostly mathematical analyses, not programming classes. I plan on doing assignments, especially in the cases you suggest, where a final exam isn't a good basis of evaluation. My reason for excluding them as part of the challenge is simply the difficulty in grading them.

[kunjaan]

@ScottHYoungVid Don't worry about grading. Here's a challenge - do the assignments and post them on the internet. Somebody is bound to correct you. However, still have the final exam as your "pass or fail" criteria for theoretical or math oriented courses. But STILL DO THE ASSIGNMENTS.

[wellsjohnston]

I'm sorry, but as a cs student, I can say with confidence that you will not understand the material well enough in under a year to pass the upper-level courses. It takes months to wrap your head around the various levels of abstraction leading up to design patterns and advanced algorithms. And that's just the beginning of courses like compilers or AI. Not to mention, you're going to have to learn digital logic, basic circuitry, and computer architecture. Nice publicity stunt.

[HeiLong24]

@wellsjohnston Agreed. I've graduated with a dual CS Math degree in 4 years and I'm still a mediocre programmer, terrible compared to my coworkers. Taking all of these courses in a year to get an idea of what CS is about is one thing. Actually understanding the theory and industry practices (which seems to be what he wants) is something that I don't even think most programmers would claim to have a firm grasp on.

[ScottHYoungVid]

@wellsjohnston - Part of the problem with running a challenge live is that it appears rather arrogant. That's an unfortunate consequence of trying to document it live, so we'll just have to see how it turns out when I cross those points. You could very well be correct, which is part of the reason I'm having a faster pace in the beginning so I can give more focus to the higher-level classes.

[wellsjohnston]

@ScottHYoungVid That's probably why my response was so cynical. You might be very capable of understanding all of the material leading to upper-level courses in a year, but you wouldn't have time to do the engineering. While learning CS concepts might not take that long, the programming actually does. My point is you need the 4 years to complete prerequisite material, practice programming, and facilitate multiple hair-pulling semester-long projects. Regardless, you're going to have fun.

[wellsjohnston]

@ScottHYoungVid That's probably why my response was so cynical. You might be very capable of understanding all of the material leading to upper-level courses in a year, but you wouldn't have time to do the engineering. While learning CS concepts might not take that long, the programming actually does. My point is you need the 4 years to complete prerequisite material, practice programming, and facilitate multiple hair-pulling semester-long projects. Regardless, you're going to have fun.

[ScottHYoungVid]

@wellsjohnston - I'm not in the field, so I have to counter with a quote from Cal Newport about this very project, who did his PhD in computer science at MIT and is an associate professor at Georgetown: "Programming is great, but it's not computer science. Asking why someone studying CS doesn't do more programming projects is like asking someone who studies mechanical engineering why they don't spend more time repairing cars." Programming is important, but it's not the main goal of MIT's CS.

[wellsjohnston]

@ScottHYoungVid That's true, CS is not software engineering, and programming is not the goal of a CS. But that doesn't mean you can learn CS without programming, just like you can't learn math without doing proofs. Read about recursive backtracking algorithms all day, but could you use that knowledge to build a Sudoku solver? Implementation skills are pretty much already expected of people learning CS. Even if programming isn't required for CS knowledge, it's a big part of the curriculum.

[ScottHYoungVid]

@wellsjohnston - I never said I wasn't going to do any programming. I'm merely making the point that for the majority (although not all) of the classes, the final exams are the significant basis of evaluation, therefore justifying my claim of focusing on the exams as the key criteria. Frankly this debate is meaningless. Either I'll be successful, or as you seem to suspect, I'll fail. Until then, there isn't really much to talk about.

[HyperxThread]

@wellsjohnston 1 person knows all!!!



[corrigantj]

@wellsjohnston "Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction." - Einstein

[nomanissafe]

@wellsjohnston You don't understand the meaning of the word "Prodigy"

[lateefishere]

Boy, you done lost your mind.

[flwrstrm1618]

also, what's the purpose of writing your own final exams? i never really got that. if you're writing your own final exams, it's like you should know the solutions. and your ability to write your exams should be proportional to how well you already learn the material.

[ScottHYoungVid]

@flwrstrm1618 - They aren't my exams. They're the actual exams used by MIT to evaluate students, and I have one for every single course I'm taking (and the official solutions for 31/33). The amount of official MIT material I'm using is quite substantial. All of MIT's free material can be found at: ocw (dot) mit (dot) edu

[flwrstrm1618]

scott, i've been following your blog for about 4 years, and this is the most excited i've been about anything you write, advertise, or just do. good luck, and i'd love to hear how your work goes.

when i first read your blog title, the feel of epic-ness was proportional to epic meal time. that's how epic it is!

[andrey20u]

that's cool, can't wait to see it ! and good luck ofc.

[guitarpant]

This is interesting b/c where I live actually has 12 month degree options which is an obvious ploy to get into the workforce faster.. It's also interesting b/c i was actually thinking about learning about technology to gain some knowledge in an obscure field.. how I see it, this is possible if you focus on nothing else.. I don't know if you have to worry about everyday stuff like work, family, children and relationships etc but if you don't, you clearly have an advantage

[ScottHYoungVid]

@guitarpant - Luckily my personal life is relatively uncomplicated. I do have a full-time business, so juggling that plus courses might be tricky, but I've done my best to put it on autopilot for the next 12 months.

[nzpacific]

This project could also massively increase the popularity of your blog. People are fascinated by this kind of shit you know.

[miguelangelwood]

Massive challenge! The load under normal circumstances is overwhelming. 4 into 1 will reveal alot about learning techniques and for the majority of us accessing the under utilised brain. I look forward to your insights and experiences. Wishing you all the best for what will be an intense 12 months.

[GMark0]

Scott: You will succeed. Completing a baccalaureate degree rarely involves operating at the extreme edges of intelligence. What it does require is focus and discipline, which you have demonstrated in spades. My only recommendation is that you keep the end in mind -- you can paint into the corners to get to 99.9%, but the Pareto principle probably applies -- 80% of the value you obtain will be from 20% of the material. Learn to discern, and the extra time will go toward creating success.

[polykow]

good luck!

[tinylittlecricket]

Sweet! Looking forward to gaining valuable insight thanks to your courage and openness.

[theadmini1992]

That sounds very intense! I wish you all the best with it. ;)