Ask HN: Makes sense to do Masters in Computer Science after 13 years experience?

Great response! RE: machine learning. Where would you say you are now in terms of expertise? Have you churned out some personal projects? Are you working on it professionally?

I would recommend to do that. In Germany (depending where you are located and the school ) I even could keep my job and just extend the study time to about double the regular time. I didn’t learn a lot of new stuff but I was „forced“ to repeat and relearn a lot I wasn’t using in a long time (math, proofs, low level electronics in my case). Doing that didn’t change a lot regarding my income but helped me to reflect on my daily work. It feels a little bit weird to be one of the oldest in the lecture though.

Edit: Education is pretty much free here, the only thing you’ll have to spend is your time.

What's your goal? If you want to learn, get an MS. The opportunity cost of leaving the industry to go to school for 2 years is extremely high, though. You'll miss out on presumably somewhere in the area of $250k of income at non-SF levels for a senior-lead engineer.

How are you going to be admitted to a graduate program without an undergraduate degree? The cases I'm aware of where that's happened are so rare and unique they may as well have never happened.

Even in those instances, I don't think any of them happened in the last decade or two. If anything it's probably more difficult than ever to e.g. impress the right faculty member and thus bypass graduate admissions these days (Lockhart did this after meeting Erdos at Columbia in the late 80s). The amount of bureaucracy has only increased over time. More importantly much of it has been systematically automated, so you longer even have a clear system for a human to bypass a requirement in exceptional cases.

I think it's also hard to make a really good case for it. I know someone who was admitted to Harvard at 16 years old. Upon admission he already had two full years of undergraduate credit completed through a gifted youth program at Stanford. He finished his bachelors degree before he turned 18 and received his PhD at 21. He didn't "need" undergraduate education in the usual sense of the word, but he still got it (albeit in a non-traditional way) before going to grad school.

It might sound harsh but when people like that exist and still get a bachelors degree, universities don't really have to make exceptions. There's just no pressure to do so, and in their eyes anyone without the qualification is an unknown quantity with an unfamiliar bureaucratic exception process tied to it.

Surely "exponential personal growth" would be an artifact of the method of measurement? Your first task could be to hack that method to get an arbitrary conclusion of your choosing. I am totally serious. Bigly lesson, that one :)

Remember how, as a kid, doing the wrong thing was thrilling? For me, that's what enjoying learning is like and it does often start from a willingness to be wrong or fail (ever ridden a bike into a bush - not as soft as you think at high speeds).

For example, yesterday I ended up randomly understanding Godel's proof for his incompleteness theorem and it's significance because I was trying to find out why the validity of substitution/equivalence seemed to be an assumption of all logics (if anyone can point me in the direction of what I'm misunderstanding there, that'd be great).

Now, almost everyone I've ever met tells me I'm crazy and/or an arrogant asshole depending on context. It does worry me but not enough to stop because how I feel is like a slow burning version of jumping off a cliff.

It's scary and hard to get started but once you do, you experience something fleeting and hard to grasp and then a sort of calming shock as you hit the cold water and look up, somehow feeling relaxed about having a long way to swim to the surface.

(Bear in mind that I'm paranoid about checking for rocks before jumping and that is somewhat analogous to learning - some of what you do is swimming around at the bottom diving down to see how deep it is or finding a vantage point where you can see the bottom from. It's also hard to trust someone when they tell you it's safe. All of that features in learning for me as well.)

One of the things that happens if you do things like that is that you get fitter because you've been climbing up to the top of that cliff over and over. If you then have to do something a bit boring like, say, read the manual for Rust, it isn't very hard because you're stronger, fast have better endurance and are less likely to trip.

What I'm saying there is that you may have started at the crappy end of the experience :) Go have some fun, find a motive and then worry about it.

Re: distributed computing... If you want to do something cool, I'd suggest starting with Aphyr's blog and his Jepsen test suite.

This is written in Clojure, which is JVM based but has a javascript version, ClojureScript.

Lots of Clojure packages work out of the box in ClojureScript and last time I had a play the "Leiningen" package manager understood that and came to the party.

From there, you could adapt Jepsen to testing a distributed system of your choice communicating via WebRTC and a STUN server running in a serverless browser environment.

If you go straight to conflict-free-replicated-datatypes (CRDTs), I think that'd all qualify as cool - especially if you can automatically test them (I'd suggest spinning up multiple browsers using puppeteer).

Remember that composition still exists and it's possible to achieve slightly less abstract programming conditions by creating a hard dependency on a reasonable context that covers some of the requirements of CRDTs.

/endbraindump glhf