freax

...but if you look at the companies he _actually_ funds, it ain't so.

http://www.techcrunch.com/2008/03/14/y-combinator-demo-day-r...

I forgot it was 2008. Cripes, I'm getting old.

It still doesn't make sense. You say two years ago you had this max-timeframe criteria; but there's a startup you funded two years ago that has yet to launch. Well, you can't use them as an example of your foresight if they accidentally slipped through your filter, or if they accidentally ended up taking much longer than expected. Fo' shizzle, G.

It's kind of amusing the rejectees not only got a form letter, but a recycled one, which wasn't even accurate!

I think what's being overlooked by the article is the actual development of some good technology -- there seems to be so much focus on the exit rather than what a startup is actually doing.

So many of these Web 2.0 startups seem to want to grab a piece of the pie by diverting traffic rather than growing the pie by offering something fundamentally useful.

From your previous letter: "We rejected a lot of proposals [...] because the project seemed too big to start on only three months of funding."

All your non-defunct fundees are still "building", but that's not the same as accepting groups you know are going to take two years and a lot more funding right at the outset.

As for caring about the founders/investing in people instead of ideas, that doesn't hold up for several reasons: 1) If you really end up changing the idea so often anyway, as often claimed, the idea proposed wouldn't be grounds for rejection if the founders were otherwise good; 2) Many founders don't have track records, thus you have nothing to judge them on; 3) Some applicants do have track records, but you reject them without reviewing the stuff they've done due to lack of time or other reasons.

> The reason there aren't more Googles is not that investors encourage innovative startups to sell out, but that they won't even fund them.

This goes for YC as well. Surely there are some technically innovative applicants that don't fit the mould of web sites to make quizzes or 10-second cartoons or other such dubious acceptees.

Google itself would have been rejected on at least a couple of grounds: project being too big to make much progress in 3 months (crawling the entire web!), and needing too much money ($15k ought to be enough for anybody!). IIRC they got not only the $100k from Bechtolsheim but also a similar amount from Cheriton.

> And yet it's the bold ideas that generate the biggest returns.

Yet YC sticks with smaller ideas...

> It's remarkable how wedded they are to their standard m.o.

This also seems to apply to YC. Small idea, $15k, 3 months, move your ass to Boston (then move again to SV, where you should have been in the first place), standard terms, no negotiation of the offer if YC accepts you.

Don Knuth wants to see your "O" face.

> But how many days must later be spent debugging issues like this? Will this time spent debugging be counted against your productivity gains you attribute to Lisp? This is exactly the sort of problem Erlang has nailed down.

http://damienkatz.net/2007/12/erlang_vm_crash.html :

"So the Erlang VM died suddenly with a failed memory allocation. This is actually quite okay [...] However, the Erlang VM didn't restart automatically. Apparently we need to configure something for that to happen, only we can't figure out how to make it work. The documentation is either wrong, confusing or the feature is buggy. This is the kind of stuff we keep hitting in Erlang. It's fantastically productive for many tasks, but using some of the built-in libraries and features can be a huge time sink. (inets and xmerl immediately spring to mind)."

> Since everyone has to obey artificial rules, there's no harm in making the rules prevent damage to the players.

You're begging the question here. Many of these drugs are overwhelmingly beneficial -- e.g. Human Growth Hormone. Prohibiting it makes about as much sense as prohibiting surgery; it's used to promote healing of injuries. If they cared about damage to players, its use would be encouraged. Reducing recovery time means they are playing and making money for the sport rather than laid up for months and just cashing checks. I would argue you have a moral duty to yourself to use any and all means to get better, and anyone who argues you should remain injured longer has no moral basis to stand on.

Baseball's rules are simply about image in this "drug war" climate. It's extremely hypocritical; Mark McGwire and Barry Bonds likely made a fortune for the league with all the excitement about breaking home run records.

It's also uncomfortably Übermensch-ish. If you disallow this type of enhancement, you're leaving it up to whoever was born with favorable genetics. It's not even about who trains the hardest -- someone with better genetics can do less work and get better results than someone else who works his tail off.

Additionally, there is a wide range of performance-enhancing substances, some of which are well-studied and can be safely taken under medical supervision. However, the well-known ones are typically the ones they test for and prohibit, which drives the practice underground and spurs development of experimental new substances of dubious safety.

And realistically, in order to compete in many sports today, you need both the genetics and the pharmaceutical enhancement. Thus the biggest threat to athletes' physical health is the prohibition of well-understood substances and the legal pressure on doctors which makes it difficult for an athlete to get the proper medical supervision. The system simply does not reward people who either lack superior genetics or the willingness to use these substances.

> the cost to rent a converter box (dual tuner HD DVR) is now $14 a month. Considering I'd need one for every TV in my house, that would effectively quadruple my cable bill!

Good Lord. I can only imagine the fearsome startups you'd lead if you could just kick that massive TV habit.

> In short, the easy pickin's have long gone, and now we need complex tools and complex coding styles to get more performance from more cores using the existing languages.

I don't think more complexity is the answer here. All this massive multithreading just makes things impossible to debug.

Why is this an either/or?

Anyway the only kickass UI is the nipple. Somehow it's just more compelling than any given layout of tabs and checkboxes....

Except brainteasers are vulnerable to prior exposure. Once you've heard enough of them you have a bunch of tools at the ready for any others that use the same kinds of intuitive leaps -- and you may not even need to do that if you get asked ones you already know.

Just pretend you're really trying hard and voila, eventually spout the correct answer you appeared to have thought up on the spot.

I don't think that assumption would hold up to statistical analysis.

> all we've seen is an application form with 120-word answers, and maybe a demo.

The reject group also includes applicants that have stuff to show that you didn't even look at.

The mistaken assumption is reading things into the statement that aren't there. I'm simply talking about having done something concrete and functional rather than only having a proposal. You're leaping to conclusions by projecting some kind of finality onto that.

Well, everyone is special, Tim.

> if you have customers, nobody cares what your pedigree is.

If you already have customers, you're doing YC a favor by applying.

> YC backs teams not ideas.

Every team works on an idea. A lot of the last batch are underwhelming:

http://www.techcrunch.com/2008/03/14/y-combinator-demo-day-r...

> AT Interview stage, from one application form how familiar do you expect someone to be once you've read 500 forms

If they're flying you in for an interview you're no longer in the pool of 500 forms. If they are asking questions that make it clear they aren't paying attention to the information you've already given them it just goes to show that your grandiose plans are fairly inconsequential to them and they're only tossing their spare change at you.

> Some companies like rescuetime and heroku were well past the idea stage

Companies that have a lot done already are the ones that are going to run into the mismatch I talked about. Here's this great thing you made; now you're giving ownership to people who don't feel about it the same way. There's not a lot of investment of money, time, or emotion on the part of YC. So it seems to me that the exchange is equity for exposure -- basically you've started off your marketing budget at the 6% that YC takes.

> but everyone thinks they must be a great product advisor

Saying PG is better than the average random person with an opinion isn't giving a lot of credit.

> 1. Product advice: Paul Graham is one of the most astute product advisers i know.

I am sure this is true, but how many product advisers do you know?

When I look at that last batch of acceptees... I am very surprised at some who made the cut. I'm not sold on YC's judgement. I think they may hit a few home runs just due to taking a lot of swings rather than due to an ability to identify the good pitches, and get lucky that a couple hundred fair pitches are proffered every time. Which leads into the next point...

> Pg and co get so many applications and have so little time to read them that you should expect that the YC partners don’t have any idea what you do

...Which means you're giving ownership to people who don't care enough about what you do in particular to spend time on it; this is a big mismatch if you've spent a lot of time on your project, so their approach is more aligned with people who come in with ideas sans code. They have dozens of other start-ups to deal with, not to mention all the time some of them spend on this site.

So the main value in YC seems to be demo day -- getting a stage to present to a large number of moneyed interests at once, plus the news coverage, so you either get the real investment dollars you need or get acquired.

All the information you get is pretty much replaceable -- you can find it other places -- but YC takes the write-protect tab off the floppy so instead of just you being stuck in read-only mode you can write out information that is seen by people you want to see it.

At least that's what it appears their primary value is to me.