Everyone seems to have a certain implicit sense that, sure enough, a CS education isn't crucial to work in certain sub-sets of software development. Likewise, it's well understood that such a foundation is necessary in others.

> For example, if you're doing anything with machine learning, you might want people who understand statistics. Oh, and if you're designing aircraft, you might want someone who's studied aeronautical engineering

Exec, of course, is neither doing cutting edge work in machine learning nor designing aircraft. Neither, if we are honest, are most startups or software companies hiring developers (including those who emphasize & prefer pedigree).

The question being posed (continuously) is :

To what degree is being self-taught (or, non-university taught) a hindrance in positions that do not explicitly call for it? If it's not a major hindrance, then is limiting the hiring pool in fact irrational?

I do not see this being answered adequately and directly. More often than not, we default to pointing out edge cases where a university education in CS/robotics/engineering/bio-engineering/what have you is necessary to perform the basic job functions. This isn't particularly productive.

Data should resolve discussions, not corner cases or anecdotes. Is there anything actionable or quantifiable that we can work with, or will this discussion inevitably lead down the path of opinion?

https://iamexec.com/meet_our_execs

the company may not care about credentials for hiring, but for marketing purposes at least they are creating the illusion that their menial helpers are in fact credentialed.

further off topic - it's a little depressing to me to see the "high tech" cradle of silicon valley creating a service to organize menial labor. wow, house cleaning online is sexy? i must be dreaming, is it 1999 again?

Regarding listing one of our Exec's degrees: credentialing clearly matters to people outside the company (customers), and there's little we can do to immediately change that. When we talk about credentialing not mattering for hiring, it's for core team employees (people who build the product), for which there are generally other metrics we can evaluate on.

i think robots picking artichokes would be cool in terms of both high tech and reducing the dependency on exploitative labor conditions. but moving robots and object recognition are tough problems, and when there's other "low hanging fruit" (excuse the pun) to be found in other startups the technically difficult stuff can get pushed off.

exec may also have longer life as a viable business than facebook. it fills a need that won't go away, whereas facebook has a major risk of having the fad end, or alienating users through ever-increasing invasion into people's personal data driven by the need to justify a ridiculous valuation.

only risk i see to exec is what happens to your quality labor pool if the job market tightened, but that doesn't seem like a big risk for a while. with 8+ million people dropping out of the labor force over the last four years, there's a lot of slack to pick up.

i agree that credentials are not necessarily indicative of on-the-job effectiveness. alternative and cheaper ways to hire people, like using programming tests (we use them at my company), are tricky and can risk running into discrimination lawsuits if they are not directly job related, esp at bigger companies. however, for some reason using tests to filter people out is considered OK if it is done through a university, and then employers hire on the back end, which leads me to think that it is partly employer laziness and partly fear of liability that keeps the credentialing system intact.

Generally, the hard problems with robotics are related to sensing. Stuff like inverse kinematics and gripper movement are mostly handled in ROS if you can build a model.

Recognising the artichokes is not that hard as you might think given OpenCV as a primitive, and if you could get a near-field sensor using structured light it would actually be easy. This is not possible right now but will be in the next few years (unsubstantiated prediction).

Anyway, what I am saying is that you would be surprised how fast the boundaries change between "hard" and "easy". The things people are doing now with a $200 irobot create and a $100 kinect are blowing my mind.

I do agree that house cleaning online isn't very glamorous, but if it helps people and makes money, it doesn't need to be.

What I really need is a decent tutor who is willing to help me specifically with the issues I have but I've sought in vain for one who is willing to free-wheel it with me instead of just copying out of a textbook. I tried one briefly and he came up with a cop-out shortly after our first lesson, because I don't think he liked the unconventional questions I was asking, like "Why and/or how are sine waves relevant to non-geometric data? The only definition I can find of a sine casts it in strictly trigonometric terms, so how is it applicable to non-trigonometric data? Is everything encoded into a representation of a triangle before these calculations are applied?" Heh, that one made him pretty annoyed and he didn't really have a good answer.

I would love to increase my background in statistics and comp sci theory (which is basic but imo sufficient, and I seem to have a better grounding than most of the CS grads I've worked with), but I don't really know of a good option to receive that training. If someone wrote tutorials for graduate-level math from the bottom principles up like they write out tutorials on PHP or whatever, I'd be all over it. I really want to increase my formal mathematical literacy.

Three answers (in case you haven't found one you like yet):

A set of sine and cosine waves whose frequency is a multiple of some value form an orthogonal set of vectors/functions, which means that for any given function or vector whose domain is at most the period of the lowest frequency wave, there's exactly one set of weights whose weighted sum equals the function (with certain caveats if the function isn't discrete). There are other such sets, for example the Hadamard series, so sines and cosines aren't unique in this regard.

A complex number can be treated geometrically, as a phase + magnitude (converted to a+bi using sin & cos of course). This representation has the benefit that the magnitude of the value is readily apparent, and makes certain calculations involving multiplication and exponentiation easier.

Sine waves arise naturally in differential equations, because they are the only functions which are the negation of their own second derivative. Hence they often turn up in second-order systems with negative feedback (e.g. microphone feedback is more-or-less a sine wave).

you can also show up to public lectures given by visiting math profs and afterwards ask them whatever theoretical background questions you want so long as it's not total spoon feeding

campus walls are also covered in tutor posters for hire and many of them graduate level

I am a mathematics professor. Where do you live, how is this advertised, what kind of audience does this attract, and what sort of questions get asked?

Here they mainly talk philosophy and crazy advanced, graduate level mathematics that are way beyond my comprehension and often there are industry programmers, visiting professors on vacation, math self taught geniuses who smoke a pipe with huge unkept beards that look insane, students and even this anarchist group that shows up sometimes to talk game theory.

A few Math dept profs hang out on Sundays here too where all the public chess boards are set up and are fully approachable to answer questions as long as they aren't engrossed in a game.

Clicking on the Events page for the UBC Math dept they always have visiting Math profs give free seminars to anybody who wants to show up, and it's easy to get to the university. Every month at least 5 seminars there's one coming up by a visiting prof from UC Berkeley on Lattice Poisson AKSZ Theory, a bunch of discrete math seminars, and 2 seminars today on chemical distances and shape theorems in percolation models with long-range correlations, and retractions of representation varieties of nilpotent groups.

These guys stick around afterwards and are fully approachable I would talk to them all the time about offtopic theory and went to the student bar with a few of them and other students for a few hours.

The universe likes sine waves. If you hang a weight from a spring and give it a yank, it will bob up and down according to a sine wave. Lots of resonating and oscillating things are also governed by sine waves. So scientists and engineers are forced by circumstance to learn all about sine waves.

This rings true to me. I have a degree in aerospace engineering, but I'm a self-taught programmer. I've worked as a programmer, never as an aerospace engineer, but to this day I have a much better theoretical grasp of the latter than the former. I was recently trying to understand bidirectional type checking, and I'm just stuck. I understand ML-style type inference based on unification, for which there are a lot of undergraduate-style descriptions online, but I'm clearly missing half a dozen classes between "point A" and "point B" when it comes to anything more advanced, and I don't know enough to know what I need to learn.

Some things are just not easy to pick up, even for people skilled at self-education. They require a level of background knowledge that has to be acquired incrementally, and it takes time and discipline to pursue that path systematically.

http://www.cis.upenn.edu/~bcpierce/tapl/ (It is very amenable to self-study)

and then:

http://www.cis.upenn.edu/~bcpierce/attapl/ (Began working my way through this, but haven't quite)

I think knowing abstract algebra is pretty helpful to understand types, but I'd imagine you were likely exposed to a great deal as an aerospace engineer already.

This brings out a point not explicitly mentioned in your original post. A key thing you get in college is access to people who know the topology of your field. It is a routine recommendation to start with TAPL. A map is invaluable in order to limit backtracking or dead ends.

In my field of reverse engineering, those who have produced the top public results are completely self taught. Advanced forms of reverse engineering require a large smattering of knowledge from many graduate level areas of Computer Science and Mathematics. Autodidacticism of some form is the only option, whether it involves complete self-instruction or designing a custom curriculum in graduate school. The aforementioned authors have described the process of learning without a roadmap as extremely painful. If you don't have to subject yourself to it, I don't know why you would.

I think that illustrates the real value of education, which has nothing to do with brand name or credentialing. I like that I studied enough electrical engineering to do hobbyist projects with FPGAs; enough physics, math, and other sciences to be able to make sense of Nature articles, as well stay up to date with relatively new and fast growing fields like neuroscience and molecular biology.

Not everyone will extract this out of their degree program and there are definitely a few universities that make it difficult to get this kind of background knowledge. However, I'd wager that most good universities (ABET accredited CS/engineering programs, most faculty having Ph.D.s and publishing, healthy portion of students going on to graduate schools, etc...) offer this to students -- irrespective of their USNWR rating is (which, I think, at least for general undergraduate studies becomes more of a game beyond a certain point).

Could MOOCs offer this? Probably, but having structure and providing a toplogical sort (just like you've described it) of disciplines -- as well as things like labs for hard sciences -- is also valuable.

==Very good point.

[0] http://www.seas.upenn.edu/~cis500/current/index.html

I wonder if some better way of discovering books suitable for self-study would solve at least a part of the problem. I find a lot of textbooks are aimed mainly at being used as a resource in a course, which isn't quite the same use-case. Others are more of a compendium or reference, which also isn't the same thing, e.g. you could learn algorithms from either CLRS or Knuth, but I don't think they'd be engaging as introductions. So far my method is to ask around and make notes when people suggest things in threads like this one.

An even rarer genre of books is the high-level, creatively written book that lets you understand why an area is interesting in the first place, but which still digs enough into the concepts that you learn something about it. Sort of the textual equivalent of the dazzling lecturers you occasionally run across in universities (alas, not most of them, myself included, though I do try to provide students with a high-level map of what's going on in an area). There are some good books in physics, which attracts a lot of good popular writing leaning toward the harder-science end of the spectrum, but not as much elsewhere. In CS, only Hofstadter's Gödel, Escher, Bach comes to mind, though I'm sure I'm overlooking something.

What Ivy League schools do teach better than other schools is exactly what Justin says: how to fit in to the elite.

Like Emmett point out, however, I think any potential incremental benefit of a CS education at a top school is very much overvalued by recruiters.

The "everyone" that you are referring to consists of the vast majority of university majors, which happen to be non-vocational. I would say that STEM majors are generally pretty close to vocational, but as you say lots of people can get the degree without actually learning anything useful to any specific employer.

Which is what vocational schools are for. The biggest pity is that vocational schools aren't held up as great opportunities - lots of them are intellectually challenging - can you name and describe the function of all of the moving and non moving parts in your car? Also vocational schools are just as social as long as they are encompassed in a normal junior college. Oh and JOBS that pay MONEY.

The funniest part of the tragedy is that (on average) someone going to vocational school for a law certificate is going to make less money than someone going to vocational school for an electrician or plumbing or automotive repair certificate. Have we mentioned debt and compound interest?

Yeah, well, you know, that's just like, your opinion, man.

Memes aside, that contrarian claim may or may not be true, but where's the evidence to support it?

On a related note, the OP lost me with the claim that "Speaking as a graduate of one, top schools [just] teach you credentialing and ladder climbing." That doesn't match what I've seen at MIT or CMU, but perhaps it's true at Yale? :P

At the very least, it's an extraordinary claim (for engineering majors) presented with something falling far short of extraordinary evidence.

I am a physics graduate of a state school. I thought my physics program was excellent. I have since studied some additional math from MIT's OpenCourseWare and some machine-learning classes from Coursera. In my opinion, the quality of the classes are roughly the same.

But what I have noticed is that a lot more of the original research comes from the big-name universities. (I suppose that professors compete to go where the most funding is, and also where they can get access to the best grad students.) If I wanted to get a PhD in machine-learning, I'd most want to go to U.Toronto or UCL (University College London). Because that's where the action is.

As a graduate student, if I'm studying the xyz algorithm, I might be learning it from the guy who invented the xyz algorithm. (As well as surrounded by grad students and post-docs who are studying the next version of the xyz algorithm.)

I suppose that, in the STEM fields, there's a trickle-down effect to the undergrads. Surely there's more energy in studying something right where it's being invented.

The comment was claiming that Justin claimed that education didn't matter. I was clarifying that his point seems to me to be that the prestige of the institution you attend doesn't matter, and that more prestigious universities do not offer correspondingly better educations.

Now, as to the truth of that matter, it's a topic there's not a lot of good evidence on.

Obviously MIT and Yale and CMU all select from the very best students, so their graduates are correspondingly better than average as well. The question is whether they are more-better after attending prestigious schools.

Anecdotally, my friends who went to state schools got educations every bit as good as mine, in terms of the quality of instruction and rigor of the classes. Ditto for me comparing my CS degree to a MIT degree (I don't know about CMU).

Also, the null hypothesis (or the dominant prior if you're bayesian) should be that schools provide equivalent educations unless you have some reason to believe that's not the case. What evidence do you have that MIT and CMU provide better educations than other schools?

Again, note that the claim is not "You don't learn anything in college", the claim is "You don't learn anything more at prestigious schools".

Fair enough.

Obviously MIT and Yale and CMU all select from the very best students, so their graduates are correspondingly better than average as well. The question is whether they are more-better after attending prestigious schools.

Sure.

[The] null hypothesis (or the dominant prior if you're bayesian) should be that schools provide equivalent educations unless you have some reason to believe that's not the case.

See, this is where the argument gets pretty flimsy. The prior should take into account the data we have on hand and practically all of it points in the opposite direction. Some data we have includes:

* professors (in many cases, learning techniques from those who invented them)

* quality of students you're interacting with

* laboratory resources available

* the "wisdom of the crowds" about the schools in question

Sorry, but the burden of proof is yours (or Justin's) — practically ipso facto — given that this is a contrarian claim.

From a pure common sense perspective, we should be surprised if there is literally no difference between the CS education at MIT and, say, UC Irvine, just as we should be surprised if there is literally no difference between the dining experience at a restaurant with multiple Michelin Stars and that at one without.

Well that's quite a bit harder to refute. ;)

For example I am in the UK and studied computing at an FE college (probably equivalent to a US community college) and also at a "red brick" university.

Both courses contained introductory Java programming modules. At university the first few lectures were spent whizzing through the Java syntax and by about the fifth lecture we were being introduced to BST implementations.

The college course took about the same time to explain the difference between an object and a class and many students still struggled.

Of course if you are comparing "good college" to "really good college" the difference in intelligence might be much more minimal and admission comes down to whoever studied the hardest for their exams.

At the time I got my engineering degree, my dept was in the top 20. It was pretty clear through the entire 4 years that educating undergrads was not a serious priority for the professors or the dept as a whole; research was king. I'd be pretty open to evidence that colleges without world-renowned research are actually better teachers.

Same here (DC' 04) but from my experience places like Waterloo produce graduates that are phenomenal compared to Yale. They leave Waterloo with such a wealth of actual working experience. Even when I was at Yale recruiters from Microsoft would complain to my professor that Yale graduates were lacking in actual experience. Don't discount the value of experience in learning theory because a great deal of CS is driven by real problems encountered. It's harder for the theoretical stuff to sink in without understanding the problem they can be applied to. I don't think I am alone in this regard.

In any case, dollar for dollar, if you're going to hire new college grads your money will go farther on graduates from schools that have a strong internship program. So in a sense, school does matter.

Maybe what is more true is the label "Ivy League" doesn't matter, not in CS anyways.

It's also not particularly prestigious, which just goes to prove Justin's point.

It might not be prestigious to some people but where I work Waterloo carries a lot of weight. We also love Brown.

I was simply appalled, when I noticed my co students in computer engineering had never seen a computer from the inside or knew that c was a programming language I studied. every single one of them has a ph.d now. I pretty much left.

And now that I moved to the US I can tell you each one of those who studied in a german engineering school is worth more than your average Ivy league candidate.

See, the thing is, the ones that actually do care drop out, because the system is made to find the highest common denominator. You want to get as many cs engineers out as you can from your university.

I often go on to tell people you should get interested high school students. Most of the ones in the University are already damaged. The ones that know coding at highschool age, are the later self taught wizzes

I was recently involved in hiring cs graduates for a university program, it was a horrible experience. They did all these capability assessments, grade measurement stuff, etc. In the end I got bored, and started asking them to solve problems(yes, like every half decent interviewer would do).

You'd be surprised how few people can break down a problem that would be 3 if statements into something that actually resembles a function. Sure, they can do exactly what you want them to do, but if that was it, you might as well just outsource it to some dude in india.

The foundational "computer science" skills held in such esteem by companies like Google amount almost entirely to an understanding of algorithms and data structures (you will almost never see an interview question based on, say, programming language theory), which was covered by exactly one semester-long course in our curriculum.

I think we learned a lot of cool stuff, but not stuff a working programmer really needs to know.