Google's first employee | Hacker News Reader

Google's first employee

mercurynews.com

140 pointsdirtyaura46 comments

I took more notes the first time Craig spoke at YC than I've done with any other speaker. Most of the time I only write down one or two sentences. The first time Craig spoke I took 4 pages of notes.

I think stories by early employees of successful companies are often the only way to see the messy truth of early stages. Founders are so involved in many aspects that constant self-observation is not possible.

At least books by founders highlight the big lows and the big highs, but there are many insights to be gained from smaller daily decisions or way of making them, and I think employees can remember better how was the daily life in the early stages.

I've found this too. We also had Shel Kaphan speak, the first employee at Amazon (and really de facto cofounder), and he was very good as well.

Reminded me of http://steve-yegge.blogspot.com/2008/06/done-and-gets-things...:

"It's modeled on the early successes I've witnessed at Geoworks, Amazon, and Google, all of whom had one thing in common: they hired brilliant seed engineers."

After re-reading the post, this sticks out prominently:

"I've been debating whether to say this, since it'll smack vaguely of obsequiousness, but I've realized that one of the Google seed engineers (exactly one) is almost singlehandedly responsible for the amazing quality of Google's engineering culture."

Is it Craig?

Dunno, some folks in the comments also mention Jeff Dean.

What made his remarks so notable?

He was the only one besides Larry and Sergey who had first hand knowledge of some of the stuff he talked about, because for quite a long time the company was just those three.

Can you please share the notes?

Unfortunately I don't think I could. If speakers knew that even notes from what they said were going to get published, they wouldn't be so candid. I talked about that here:

http://ycombinator.com/atyc.html

What I do do though is incorporate their insights into the general advice I give in essays. That's not just based on my experience with YC-funded startups, but also on things I've heard in talks at YC.

"A There are these cultures you see in some companies -- they are just very negative, where the way that you show you are the hot shot engineer is by showing how stupid everyone else's idea is, or how bad everyone else's code is. And honestly, we see that more at Google now then we used to when it was small. And part of it is we've got people who were used to these other cultures; it's a very prevalent, sort of alpha-dog culture you see in the programming world. But it's never been part of Google's culture. And I think even when you see it now, it's not as respected. The respected voices say, hey cut it out. "

--That's very true. What I have seen in large companies, looking good is important, and apart from building stuff and bragging about it, and even easier way to do it is by criticizing others, and finding any way on how to point at minor flaws on other's people's work, simply malignant / non constructive behaviour, but masqueraded in the name of 'quality', company standards etc..

Yeah, I think part of how google does it is you have people like Guido who are super humble and helpful to everyone. That was generally true of the smartest folks at google and it just set a really good example; people weren't afraid to "look stupid".

I think this is changing.

Corporations are not the blind, mindless creatures that people seem to think. They do see their own problems, and do want to change them. They are starting to recognize the needs and desires of a younger workforce, and changes are starting to occur. They also have been hard hit by the recession and cannot waste their energy on their own politics.

I would predict that in 10 years, corporate culture will be more like the startup cultures of today, but with some added discipline to meet the risk tolerances and business continuity needs of a large company.

Here's a very tiny "company" anecdote. I'm volunteering on a remake of an old game for fun. So far I've been the only programmer. I've had conversations with two other programmers who considered joining, and the conversations went similarly. That they seemed eager to tell me about something I was doing wrong and needed to do differently. I'm not really against that (even less if it comes with a patch...) but I guess it was the delivery that got to me.

For instance, in one conversation the first thing the other programmer said to me (after chatting with the guy introducing us) was "I've found tons of bugs in the build." My initial reaction was "well duh" :) But it didn't really set a good tone for the rest of the conversation. Haven't heard back from either of them. But it got me thinking in part about why I had such a negative reaction to it, but also what draws people like that. I figured it was because it's a game, but after seeing this I guess it's common in the industry (I'm not sure if I should be happy or sad).

Perhaps it's because we're working in relative anonymity on this game and a large corporation has a similar feel that it still would remain only there. I hold out hope.

I'm not sure if the parent link works or not, it didn't work for me.

I believe this is the article linked.

http://www.mercurynews.com/portlet/article/html/fragments/pr...

Edit: Nevermind, now it wants to work. Although I did link to the print version, incase anyone wants that instead.

I had problems with it too, but got this working: http://www.mercurynews.com/ci_16291970

Yes, you need to drop the query string from such URL's. This has been the case with at least one other recent Mercury News post.

Odd, yeah site seems to require registration with a random(?) pattern. Maybe someone with enough karma can edit the original link to point to printed version, seems that I can't.

If you append a source param to the URL it will always work (I'm familiar with the CMS):

http://www.mercurynews.com/ci_16291970?source=tupac4eva

I started here at Google NYC not too long ago, and was surprised to find Craig inconspicuously sitting at a desk across the room, tapping away at his keyboard. Some folks around here don't even know who he is.

Silverstein's old Stanford homepage is here: http://www-cs-students.stanford.edu/~csilvers/ Interesting to get some idea of what his background was when he went to Google.

I remember one article of Steve Yegge in which he wrote that for the most part the engineering culture at Google came from an early employee. Is this actually the guy?

He would be my vote.

One of the authors of the MapReduce and BigTable papers would be my guess: Jeff Dean.

He got 1600 (perfect score) on SATs.

I wonder what the relationship of SAT scores and highly successful people is.

>I feel a lot of our success is due to luck. I guess what I'm most proud of is successfully keeping the culture as well as we have, given all the success and growth that we've had.

Suggests that he wouldn't believe there to be a direct correlation.

Most smart people I know got a 'good' (say, better than ~1400ish) score on the SAT. I can't tell the difference between someone I work with who got a 1450 and a 1550 though (unless they tell me, of course).

Most smart people I know also got a better than ~1400ish score on the SAT, but a disproportionate number are also in the near perfect range (~1570-1600). I can definitely tell the difference between a 1400 and a 1570, and can usually tell the difference between a 1570 and 1600.

I find myself very skeptical of your last claim. According to the research my friends and I did when taking the SATs, the difference between a 1570 or 1580 and 1600 was often a single question.

The scoring scale varied a bit based on the difficulty of the particular exam you sat, and as we were aiming for scores near the top of the scale we hoped to get an exam that allowed a question or two wrong to still get a 1600. Often one question wrong on verbal would still score 800, math was less forgiving of a single mistake.

I have trouble believing there could be any distinguishable difference in practice between the population of students who got two questions wrong on the exam and those who got one wrong.

There's fairly little difference between someone who gets two questions wrong on the exam and someone who got one wrong. There's usually a fairly large difference between someone who gets two wrong and someone who got them all right.

On the modern (post-1995) SAT, 1600 is a fairly wide range, from people who got 1-3 questions wrong to people who breezed through the whole test, double-checked everything, and didn't miss a single question. If you're talking to one of the latter, you'll know it.

You are right.

According to Wikipedia, SAT 1600 (post-1995) is 99.93 percentile, so about 1 in 1429 (3.4 standard deviations).

Given that there are millions of people taking SAT, in absolute numbers there should be plenty of people hitting the test ceiling.

Interestingly, pre-1995 apparently had much higher ceiling, with only 1 in 142,857 getting score above 1580 (4.6 standard deviations).

Which BTW means Microsoft was founded by some really crazy dudes.

Exactly. This is what Craig means when he says he got a perfect score when it meant perfect -- he got all the questions right.

For example, I took one of the SATs of that era (before antonyms were removed), missed only one question on the math section but still got a mere 770. I got over it.

This just seems unlikely. The difference between 100 points at the top end of the scale is minute -- a few problems swing -- and the change in a 1570 to 1600 is quite literally an issue of missing just a couple problems in either section.

The SAT is a horribly noisy measure of excellent in either verbal skills or math, much less actual intelligence. I can tell you I managed to never peak over 700 on the math section only to get a perfect score on both the Math IIC and GRE Math sections (much harder tasks).

The Math IIC is easier than than the SAT-I-M, but requires more subject knowledge. They test different things: the SAT-I is primarily a speed test, measuring how quickly you can reason under pressure, while the SAT-II is primarily a content test, measuring how much you've managed to cram into your brain before taking it.

If I were to send you my resume (sans test scores), do you think you could guess whether I got a 1570 or a 1600? Or is it something you can only tell from personal interaction?

It's from working on a project with someone or sharing a class with them. If you have a casual conversation or quickly glance at a resume, it's quite easy to miscategorize. But if you're working with someone day-in-and-day out for 3 months or so, you figure out pretty quickly who knows their stuff and who's just pretending.

I suppose it's possible that someone could know their stuff and yet have bombed the SAT (if, say, they went to a rock concert the night before), or they could've aced the SAT and through sheer lazyness prove to be utterly useless. But the skills that make you do well on the SAT - quick reasoning, good short-term memory, a decent-sized store of background knowledge, and an eye toward detail that makes you go back and double-check your answers - tend to make them perform very well at a project as well. That's why people use the SAT: it may not be intelligence, but it correlates very well with it.

So someone who only got a 1570 is "just pretending"?

Speaking as someone who got higher than that and still feels like I'm "just pretending" on a regular basis, yes.

Are you serious? If you did that well on the SAT, surely you see the fallacy implicit in your comment. But just in case, I'll spell it out - I can't imagine any justification for the assumption that the ordering of people by the degree to which they "know their stuff" is the same as the ordering of them by SAT score. I'm sure there's a correlation, but what you said requires something far stronger. You may not have meant it in such a precise sense, but even in that case I think my objection actually still applies.

Not all successful people have very high SAT scores, not all people with very high SAT scores are successful, but I would guess among very successful people there will be disproportionately many people with very high SAT scores.

Anecdotally, Bill Gates and Paul Allen come to mind (I think one had perfect score and one just a tiny bit less).

Also Drew Houston of Dropbox put his perfect score on Y Combinator application (which pg put as an example of successful application despite Drew having dreaded single-founder handicap :).

There's probably a correlation, and it might even be significant, but I doubt that it's particularly strong.

I wonder if there'd be a stronger GRE*success correlation?

The SAT correlates well with how well you do in your freshman year, but it's not a(s) good predictor of lifetime success.

It really depends what you mean by highly successful. I doubt that there are many successful mathematicians that missed more than one or two questions when they took it. I also doubt that there are a lot of professional athletes with perfect scores. If what you mean is "rich" then, well, athletes make a lot more money than mathematicians.

Test scores don't identify successful people, but our system is certainly set up to make it easier on those who do well on tests. I first noticed this as a freshman in college when (because of AP exams) I moved right into deep computer science while my classmates had to spend time on general education requirements. When I interviewed for my first internship, I knew algorithms and data structures while everyone else only knew basic sorting. Once I got to my internship, I was getting paid $3 more an hour than my colleagues because of the additional course credit I received from AP tests. What's hilarious is that someone who went to MIT with the same AP scores would actually have gotten paid less than me because MIT accepts almost no credit...

You can see how the above could snowball into a huge advantage upon graduation from college. Any correlation you notice between high tests scores and successful people is most certainly a result of this property of our system and not from tests themselves. The tests could be targeted towards an arbitrary skill set and those who did well would still be more successful because the system values those who do well.

Firefox says the connection is untrusted and refuses to open the page.

Just refresh, it seems to work and not work randomly.

Reply on news.ycombinator.com