Sorted by New

Wiki Contributions


Ahh, so that's why companies want you to be "aligned with their mission" so badly.

Right. Between search costs, orientation time on the codebase and business, mentor/team lead time spent teaching, inevitable moderate fuck-ups, and general odds and sods, a good junior developer at a pure software company is accruing negative value their first year, and is net negative until nearly their second.  A bad junior developer is negative EV until you can figure that out and fire him, which can easily take over a year.  Additionally, junior developers will, more often than not, leave around year four, due to desire to learn something best learned elsewhere, vesting cliffs, or just life's inevitable changes. If this sounds like a horrible investment, the upside is senior developers. In the right business they're pure money-spinning machines, and they need junior developers around. The most valuable thing a junior developer can produce is their future selves as a senior. 

The difference between a one-year and a two-year dropout is material (decreasing the flight risk), but diligent and productive time spent as founder will certainly fix that.  Don't half-ass that. Even if your startup fails (which as you note is likely), you need to end up with connections, skills, and stories.

Do ace your Calculus III, though. If the pandemic has taught us anything it's that most of humanity fundamentally doesn't get differential equations and as such can barely think about a large range of situations. You certainly don't want to be one of those folk.

In your value estimation, a UIUC CS dropout-turned-founder has about the same EV as a UIUC CS grad, but with significantly higher variance. To what extent is this view the consensus among those responsible for evaluating software talent? Would many of your colleagues instead see the dropout-turned-founder as lower-EV (or maybe even higher-EV), but don't give much consideration to variance?

You've got it backwards.  You want hiring folk thinking deeply on variance, because variance means there is a sizable upside tail.  Finding overlooked young developers in the upside tail is How We Win. Admittedly, the variance also means we also end up looking at folk in the downside tail, and your circumstances mean that we have to be even more careful than usual about those risks.  These risks include beauties like "has he ever coded at all?" (Yes, we get applications for 100K junior developer jobs from people who have never opened a text editor), "Does she think this job is beneath her?" (good lord, you'd be surprised how often that happens, particularly from new grads "from Boston"), "Have they ever shown any capacity to ship production software?" (This one can catch academics, who may very well be geniuses but have trouble shipping at tempo and with quality.), and "Is there any chance they will stay for two years?" (That's about the break-even point for new-fledged junior developers. It's also where I would personally be most concerned about you, given your dialog above.) People looking to hire junior software developers have to start by filtering downside tails, with the knowledge that if they miss one that's a six-figure mistake. But, if your hiring process is good enough to filter those downside risks, then you really want variance, because it gives you a chance to find overlooked alpha in the form of excellent young'uns.  If you do represent overlooked alpha, trust me when I say we are very primed (and very often incented) to notice it. We call it "moneyball" for a reason.

To someone who knows that universities operate upon selection effects, getting into a selective college is almost as impressive as graduating from it, but most people subscribe to the mythos that the degree is what matters (as illustrated by the sheepskin effect), which might mean that talent evaluators that read LW favor the dropout more than those who don't

I'm a fan of this community, but they haven't cornered the market on seeing past the common wisdom. My fallback way back when I was in your position was to go start coding in the bond pits on LaSalle Street. You talk about Immoral Mazes at FAANG, but that ain't nothing. Those traders were about one step more pleasant than mafioso, and I wasn't doing anything but making some numbers bigger for them. That said, they were extremely good at seeing talent and paying it exactly as much as necessary to keep it.  Good software startup folk (not just developers and managers but HR as well) know this as well. If they don't see "two-year UIUC dropout + founder" as someone they might be interested in, then you really don't want to work for them. 

(Disclosure: at no point in this have I indicated that I am not interested in recruiting you. I may read as weird-but-benign-and-maybe-even-wise greybeard, but do not assume complete altruism on my part.)

>This information is absolutely priceless to me. Thank you so much!

Don't hesitate to ask me anything.   

>I know a guy majoring in data science at Carnegie Mellon that says it's pretty glum there as well.

On an hour-by-hour basis my time at UIUC was about as enjoyable as my time in jail, but that's admittedly a question of four years vs. twenty hours.  My best friend went to CMU, and he concurs. I mentored him straight out of college, was best man at his wedding, and he's now a CTO at a 120-person startup where I work for him. 

As someone who evaluates and mentors young software engineers as part of his living and has for about thirty years now, "dropped out of UIUC to found an interesting startup" would strike me as being good a resume' on average as "fresh CS grad from UIUC".  What I would be worried about is the variance. I would grill very hard during the interview cycle, and skip no steps. The upside of the first candidate is at least as high as the second (probably higher), but the downside risks (e.g, that they are just a bullshit artist or a prima donna) are higher. Hiring young software engineers is very much a matter of expected value calculations ("moneyball", in the trade), and negative-EV engineers are very real possibilities.

That said, I very much wish you well. I've counseled three young engineers in my career to go pro without finishing college. None of them took the advice, but I'm pretty sure two of them would have had better outcomes if they did. In my experience, UIUC is a uniquely horrible environment, so much so that flight from there should always be a consideration. 

Interpersonal Entanglement

is there any aspect of human existence as complicated as romance

Yes. Parenting and politics. Given a good enough model of humanity, you could probably prove that romance comes in precisely third after those two. Unlike romance, it's not even all that sensible to consider those two with non-sentient NPCs, a sign of their inherent complexity. Otherwise, good argument.

I'm coming in late, but I will say that you should probably examine the game-design literature. They are (for good commercial and aesthetic reasons) pretty much in line with your theory of fun, and in some ways advanced of it.

Recursive Self-Improvement

The problem, as I see it, is that you can't take bits out of a running piece of software and replace them with other bits, and have them still work, unless said piece of software is trivial.

The capacity to do in-place updates of running software components dates back to at least the first LISP systems. Call it 1955? Modern day telephone switches and network routers are all built with the capability of doing hot upgrades, or they wouldn't be able to reach the the level of uptime required (if you require 99.9999% uptime, going down for 30 seconds for an upgrade ruins your numbers for ten years). Additionally, those systems require that every component be independently crashable and restartable, for reliability purposes.

Ethics Notes

The best solution I've seen to the "nuke in New York" situation is that the torturers should be tried, convicted, and pardoned. The pardon is there specifically for situations where rule-based law violates perceptions of justice, but acknowledges that rule-based law and ethics should be followed first. The codification of the rule of pardon seems to conflict with the ideas of "never compromise your ethics, not even in the face of armageddon" that you are apparently advancing. Thoughts?

Ethics Notes

Because I would give odds around as extreme as the odds I would give of anything, that if you tell me "the AI you built is trying to deceive yourself", it indicates that some kind of really epic error has occurred. Controlled shutdown, immediately.

Um, no. Controlled shutdown means you are relying on software, which should be presumed corrupted, unless you are very sure about your correctness proofs. What you want there is uncontrolled shutdown, whether by pulling the plug, taking an axe to the CPU, shutting down the local power-grid, or nuking the city, as necessary. Otherwise, Hard Rapture.

Dark Side Epistemology

I'm looking for Dark Side epistemology itself - the Generic Defenses of Fail.

Relax. It will be over soon.

We're past that now.

X is supernatural.

X is natural.

You're correct, but it will make people uncomfortable.

You're smart. You should go to college.

Dark Side Epistemology

Everyone has a right to their own opinion. When you think about it, where was that proverb generated?

In the words of the great sage Emo Phillips, "I used to think that the brain was the most wonderful organ in my body. Then I realized who was telling me this."

Load More