"The difference is that Linus Torvalds, Theo de Raadt, and RMS have all accomplished things worthy of respect. Ted Dziuba's own accomplishments are at about the same level - or *less* - than the folks he routinely slings shit at." - Jonathan Tang
"Not really. In general, it's two completely different ways of thinking. Physics is all continuous math (calculus, differential geometry, vectors, etc.) while CS is all discrete (sets, logic, finite state machines, etc.) There were two things I got out of physics that really helped me in CS: 1. A respect for data. In physics courses, they really beat into you that empirical data reigns supreme, and when your preconceived theories don't fit the data, *your theories are wrong*. That's really useful when you're tracking down a bug or profiling a performance problem or doing a usability study. Comes in really handy in general business & entrepreneurship as well. 2. A quantitative mindset. In physics, it doesn't cut it to say "Well, it looks like this experiment should look something like this." You calculate *exactly* the results it has. That sort of precision is like the difference between coding and talking about coding." - Jonathan Tang
"Mine was almost all theory, but that's largely because I went to a liberal arts school. I don't think you should *avoid* a CS degree, but I don't think you should go out of your way to get one either. It's not going to hurt you - but realize that most of what "real" software development consists of, you don't learn in school. Most good programmers seem to have CS degrees, but that's because good programmers tend to be really passionate about programming, and if you're really passionate about programming, the most fun major to spend your university education on is usually CS." - Jonathan Tang
"I was a physics major and then switched to CS in my last semester. I would've ordered the courses from the two majors as, hardest to easiest, as: 1. Quantum mechanics (physics) 2. OS Design (CS) 3. Computer graphics (CS) 4. General Relativity (physics) 5. Intermediate mechanics (physics) 6. Thermodynamics (physics) 7. Intermediate electricity & magnetism (physics) 8. Compiler Design (CS) 9. Algorithms (CS) 10. Machine Architectures (CS) 11. Intro E&M (Physics) 12. Data Structures (CS) 13. Intro Mechanics (Physics) 14. Intro CS So yeah, physics was a bit harder, but there's some parity in the upper-level electives." - Jonathan Tang
"Amherst College had you write both, assuming you took those electives. The compiler had some scaffolding, but you were compiling from a variant of the C language to SPARC assembly. The OS was from scratch, but ran on a virtual machine to simplify debugging. BTW, for the college- and high-school-age bystanders: there are 3 upper-level electives in most CS courses that you should really make it a point to take. Compiler Design, OS Design, and Computer Graphics. They are usually among the hardest courses on campus, unless you're a physics major. My boss at my first job (gap year :-)) told me to take all 3, I did, and it was probably one of the best decisions I made in college." - Jonathan Tang
"Of course actual knowledge of a domain is valuable. But actual knowledge in computers: 1. Has a shelf life far shorter than the average college education. 2. Is often tied not only to the particular subfield within software, but to the particular *organization*. I remember remarking to my boss at my first job (before college) that "I ought to learn COM, because I want to make cool Windows programs and plugins." He told me, "Why bother? By the time you get out of school, it'll be obsolete." He was dead-on: not only had COM been replaced by .NET, but desktop software in general was a shrinking field that had largely been replaced by the web. My first major project out of college was a Netbeans plugin; within 8 months, my boss called me in and said "We can't sell anything that doesn't run in a browser, so we're going to have you doing webapps." When I take a new job or start working with a new organizations, my biggest task - by far - is always learning their existing codebase, internal..." - Jonathan Tang
"Depends on the intended use of the language, but some common ones I like: * For web programming: write a Reddit clone or to-do list. * For graphical programming: write Tetris or Pong * For socket programming: write an echo server, telnet client, or webserver * For GUI programming: write a graphical frontend to your favorite command-line utilities. Most of these err on the trivial side, but they're good for learning." - Jonathan Tang
"Often you find the right problem by building your solution to the wrong problem. Build your shitty pot, show it to the client, they tell you they want a nutcracker, and that's when you build a statue. If you'd asked what they wanted, they probably would've said a candlestick." - Jonathan Tang
"> Is it actually possible to fully anonymize file origin? That was the reason for the FreeNet comment - every node on the path from source -> searcher mirrors the data, and so just because you have the bits on your hard disk doesn't mean they're you're bits. Every person who touches it has plausible deniability, the act of searching for it just spreads it around further, and you never really know where it came from. Kinda like dog shit. When I last tried FreeNet, it was utterly unusable - you had to look things up by MD5 hash, and the latencies were terrible. But that was back in 2000, and it seems much better now. So if you could fix some of the remaining usability issues and/or repurpose it so they don't matter, you could have a winner on your hands." - Jonathan Tang
"Build an encrypted P2P filesystem that anonymizes file origin and has decent search capabilities. Then release it on the main Reddit. All the folks that believe the U.S. is turning into a fascist state will go nuts over it, and poof, insta-fame. Alternatively, make Freenet usable." - Jonathan Tang
">"YC selected startup -> startup is interesting" does not imply, "Unrelated to YC -> startup is not interesting" I never said that. I said that based on probabilities (which is all TC - or anyone who deals with a lot of startups - has to go on), the chance that a startup is interesting *given* that it is funded by YC is significantly higher than the chance that it is interesting *given* no other information. And I gave numbers to support that. That's what I mean by Bayesian reasoning. Successful non-YC startups don't invalidate that. (I've worked for some of them too, at least given the criteria of success we've been using here.) But there are many more non-YC startups than YC startups. And so, if you have a startup in front of you and know *nothing* about it other than that it's funded by YC, the odds are that it's interesting are significantly higher than if you have a startup in front of you that you know nothing about." - Jonathan Tang
"I'm not going to bite on the rest of your post (really, if you don't want to get into an argument, then don't), but I'll answer this: > If I may ask, what exactly is your affiliation with YC? I've been turned down 4 times by YCombinator. I was asked to be employee #2 at one SFP07 startup. I consulted for a week for a SFP08 startup, and was offered a cofounder position at them. I turned down both, the first because I couldn't abandon the cofounder of *my* startup at the time, the second because I felt I wasn't a good fit with the other founders. I've also talked with the founders of other YC startups, both by e-mail and in person, most recently...yesterday. Everything I've posted here is public knowledge - really, if it weren't, I wouldn't have posted it. In particular, the 20% is independently verifiable: Infogami, Reddit, ClickFacts, Zenter, Auctomactic, Parakey, Anywhere.FM, and Omnisio have been acquired; Xobni and Scribd have had substantive acquisition offers that they've turned..." - Jonathan Tang
"I think there's been a lot of flack lately because a few articles from recent YC founders have come out where they talk about working 16 hour days. I also think those guys were exaggerating. Tony Wright (RescueTime founder) has an interesting post where according to RescueTime's data, *nobody* in the WFP08 session worked more than 10 hours/day of straight coding, with a typical amount being 3-5 hours of coding and 5-7 of actual work: http://news.ycombinator.com/it... I did work at least one 16 hour day when I consulted with a YC company (10 AM -> 4 AM), but it's not like there weren't breaks in-between. I think some of the YC founders writing articles figure "hey, I'm working from when I wake up till when I go to bed" and count that as 16 hours, when they've been checking blogs/Reddit/news.YC, cooking meals, going to the gym, going to parties, etc. in between." - Jonathan Tang
"I didn't downvote you, but: > Why does anyone bother reading TechCrunch when half of their write-ups are bought and paid for by the likes of pg? They are not "bought and paid for" - neither YC nor YC startups pays TechCrunch for coverage. Rather, TechCrunch covers them because they are statistically more likely to be interesting to readers. If YC has funded 80 companies and 20 of them are interesting enough to get bought out or become profitable (pretty close to actual numbers, BTW), then the chance that a given YC startup is interesting is 1/4. Meanwhile, if 10,000 companies e-mail Arrington for coverage and 100 of them are interesting enough to get bought or go profitable, the chance that a random startup is interesting is 1%. No brainer there; it's simple Bayesian reasoning. > What does piss me off, however, is being inundated with brain damaged press releases describing that new "revolutionary" web 39.2.14 Gamma invitation blagocasting service that Billy over at YC launched last..." - Jonathan Tang
"Not really. I rarely worked more than 8-8.5 hrs/day at my first job out of college, and sometimes as little as 7, a financial software startup. I got a little teasing for it sometimes, but considering that the owner tried to hire me back twice after I quit, he couldn't have been that beat up about it. Companies are very reluctant to fire productive workers, even if they work 4 hours a day and then go home. They have far too many workers that spend 10 hours/day there, produce nothing, and take up the time of those who *are* productive. Just don't expect much slack if you *don't* produce - the guy who spends 10 hours/day at work and doesn't produce gets fired before the guy who works 4 hrs/day and does, but the guy who works 4 hrs/day and doesn't produce is gone before both of them." - Jonathan Tang
"Because the money is not the point. Many YC founders don't need it. Several even have profitable companies before joining YC. YC's main benefit is *they get people to pay attention to you*. TechCrunch gets literally thousands of startups begging them to do a story on them - and yet YC startups have to work extra hard to *keep* from being outed by TechCrunch before officially launching. Try sending in your unsolicited business plan to a VC. They won't fund it. Yet YC startups present to nearly every investor in the industry at once. Of the startups that seek VC funding, nearly half get it. I think anyone who's actually been through the program would say that the 6% was well worth it. It was kinda ironic to see the article say "But I have yet to see a coder escape and not turn around and punish the next generation", because that's *exactly* what PG is doing - escaping and then helping the next generation escape." - Jonathan Tang
