> Q: Hiring?
> A: One of our rules was, we don’t want to hire your friends. Another rule was not to hire people from “lesser” universities. Another rule was to only hire people with good GPAs. It was frustrating, but it meant that we ended up with a lot of really smart people from great universities, and that served us well.
I wonder what led to Google changing their minds and not considering the GPA as a part of hiring decisions.
This is my guess. Recently Google began to aggressively recruit at my university over the last two years or so. The cynical question is, of course, "why only in the last two years?" when our particular CS program has been around since the 80s.
The recruiter claimed that it was because more people started to be hired out of my school. But that still isn't sufficient to discount this possibility.
Granted I'm living in a country that isn't the U.S., that could have played a role. But I'm not confident either way.
I believe this was done after taking a look at GPA and post-hiring performance at Google. This was not mentioned in the OP, but I guess they realized that there wasn't any or, maybe in certain positions (non-research/entry devs/etc), a negative correlation between the GPA and performance.
They looked at the data. In Work Rules!, Laszlo Bock (head of People Operations at Google) writes:
« So in 2007 we started looking for themes across the ten thousand or so people we had hired, as well as the millions we didn’t. In addition to testing technical hires on their engineering ability, we realized that there were four distinct attributes that predicted whether someone would be successful at Google:
1. General Cognitive Ability. Not surprisingly, we want smart people who can learn and adapt to new situations. Remember that this is about understanding how candidates have solved hard problems in real life and how they learn, not checking GPAs and SATs. »
The three others are Leadership, "Googleyness" and Role-Related Knowledge.
This is filled with punctuation errors. When I saw the first one, I just regarded it as a typo and kept reading. But it wasn't a typo, which is annoying—it's as if this interview was done by someone allergic to question marks.
I think what happened is these are notes some person took during a class in which Eric Schmidt was being interviewed. So the questions and answers are whatever the notetaker was able to get down in real time. The entire content likely paraphrases only the most salient things being said.
So this material was valuable enough to blog, but not valuable enough to proofread? I don't follow that logic. I can't imagine posting writing or an interview that hasn't been proofed. I think an interview with Schmidt would demand at least a first-pass edit.
Your logic fails. I should proofread someone else's blog post because they didn't, or else I shouldn't complain? Mmmmm. Got it.
You obviously don't even understand what writing is, on the internet or anywhere else. Writing is done as a service to the reader. If you can't be bothered to do an edit and correct obvious spelling, grammar, and punctuation errors, you're telling your readers that what you've written isn't very important, interesting, or worth reading.
Noam was a very successful early googler. He was one of the feel people at early Google who actually did fundamental work in machine learning (George Harik was the other).
"I went to Novell under the mistaken goal of being a CEO. I didn’t do the due diligence, and if I had, I wouldn’t have gone. Our basic goal was to get out with our professional reputations intact and not end up in jail. The books were cooked, and people were frauds. But it turns out you can overcome that, and the skills I developed helped at Google."
I'm quite surprised that he would say something like that in public. I understand that this is a transcript
made in a class in which Eric Schmidt was being interviewed.
I know the chances are slim, but can anyone confirm that he really said this?
As far as his reference to Google, it seems the next sentence is what he means: “I understood the rule of cash, so we did everything we could to manage for revenue, and the rest is history.”
That's how I understand it. I just think what he said in the sentences before is potentially slanderous. Doesn't he basically accuse Novel's management at the time of fraud?
Thanks a lot. I almost forgot about Eric Schmidt while skipping through the others. I'm currently watching CS183C: Blitzscaling-9/24/2015 with Sam Altman [1].
Well, the point of antitrust law is to prevent cartels and monopoly abuses; companies of all sizes can be guilty of such. My guess is that he's saying that EU law tends to crack down on large companies even if they are only the largest (not single) player in a competitive market, while leaving alone the SMBs that collude to corner a small market.
While IANAL, the monopoly part is a misnomer. It is really about abuse of market position.
When Microsoft got into hot water during the browser war with Netscape, it was about abusing their position in one market (computer OS) to gain a position in another (web browser and web server). This by bundling Internet Explorer for free with Windows, that in turn had special binding to their IIS web server.
So when EU is looking into Google, they are looking into how their position in one market gets used to affect other markets (like say when they re-jig their search algorithms).
> 6 months after we went to auctions, we had to merge three different databases. I asked if I should be in the datacenter, and the engineers said, what are you talking about, we never go to the datacenter.
Hah! Good one. The best datacenter is one you never see, and never have to see.
However it does make me wonder if Schmidt knows less about software than I thought he might. The random scattered mentions of machine learning present the same image. Then again, I assume this is a person taking notes which are snippets of conversation potentially out of context.
Eric Schmidt was one of the authors of Lex (not the original version) when he was an intern at Bell Labs.
He has a deep understanding of software. I've met him, he asked about my project, and his level of detailed questions showed he understood a great deal about software.
> The teams are far larger than they should be. It’s a failure of architecture — the programmers don’t have the right libraries. I hope that machine learning will fix that problem.
This may have been a transcription error, but as written I can't make sense of it. Can anyone else?
It is now understood that products can be built with far fewer developers, who leverage machine learning on large data sets. Much less code has to be written, and the ML does most of the work in tuning the product behavior.
Sounds good to me! Where can I learn about these techniques. I've got a 4 CPU Ghz machine sitting waiting for my keystrokes. Please tell me how I can get this thing to help me write software.
Hrm. So the revolution will be closed source? No, seriously... are there books on the subject? Open source libraries? Which companies are developing such tools? Googling is not turning up a whole lot.
I'd say that Facebook, Google, Twitter, and a few others are really good at this.
Many of them release a fair amount of their code as open source, but these sorts of capabilities are generally dependent on having absurd amounts of data, which is prohibitively challenging for anybody who doesn't run a website that sees massive traffic.
Are we talking about the same thing? I was referring the point in the interview when Eric says that ML will help make software development teams smaller.
I was imagining some bit of AI that had figured out the best way to download asynchronously. Or the best way to layout a UI with code. You could maybe describe a process using natural language & it would just create a best attempt for you. You could then modify or steer it.
Seems doable, but I don't see how you'd need big data from massive numbers of users. And seems very interesting given that APIS and what they can do is much more finite than the physical world.
Finding the best layout of a UI using a machine learning algorithm is an example of what I meant. In theory it means you don't really need a UX team or a UI team alongside an engineering team.
I went back and re-read the interview and I believe I'm channeling Eric properly.
I'm very much interested in these things. To remain competitive indie developers and the open source community will need to apply these methods as well. I hope to see some publicly available tools soon.
The world is biased towards good storytellers. I highly recommend reading Lauren Rivera's Pedigree. It's published in 2015 and resonates with my friends in consulting and finance.
Does the book teach you how to be a good storyteller?
BTW, thanks for this one [1]. I'll be reading it, and similar material, really soon (had it bookmarked here.)
Regarding my comment, I should have explained myself better in any case.
I suspect all software developers have experienced some amounts of stress in getting things done on time. My problem is that I don't keep a track record of it.
Just a few days ago on a phone interview they asked me of any interesting problem I had solved. My honest non-impressive answer was "nothing in particular, I just go at it and keep going".
I have faced countless problems and when I do: I fix them. I don't write them on my diary. But now I know I should have.
One thing I do keep track are my programming and linux notes because I've dealt with so many different technologies.
And will be moving those to github and blogs since nowadays you need to be a social coder.
Yes, the book helps. She wrote it to expose unfairness in American society, but unfortunately some people read it as a "how to break into the elite." One section explains how storytelling works in an interview context.
49 comments
[ 5.1 ms ] story [ 103 ms ] threadI wonder what led to Google changing their minds and not considering the GPA as a part of hiring decisions.
The recruiter claimed that it was because more people started to be hired out of my school. But that still isn't sufficient to discount this possibility.
Granted I'm living in a country that isn't the U.S., that could have played a role. But I'm not confident either way.
« So in 2007 we started looking for themes across the ten thousand or so people we had hired, as well as the millions we didn’t. In addition to testing technical hires on their engineering ability, we realized that there were four distinct attributes that predicted whether someone would be successful at Google:
1. General Cognitive Ability. Not surprisingly, we want smart people who can learn and adapt to new situations. Remember that this is about understanding how candidates have solved hard problems in real life and how they learn, not checking GPAs and SATs. »
The three others are Leadership, "Googleyness" and Role-Related Knowledge.
You obviously don't even understand what writing is, on the internet or anywhere else. Writing is done as a service to the reader. If you can't be bothered to do an edit and correct obvious spelling, grammar, and punctuation errors, you're telling your readers that what you've written isn't very important, interesting, or worth reading.
Noam was a very successful early googler. He was one of the feel people at early Google who actually did fundamental work in machine learning (George Harik was the other).
I'm quite surprised that he would say something like that in public. I understand that this is a transcript made in a class in which Eric Schmidt was being interviewed. I know the chances are slim, but can anyone confirm that he really said this?
but that's trade secret!
no seriously, I wonder if that's been tried.
http://www.theatlantic.com/business/archive/2011/10/schmidt-...
[1] https://www.youtube.com/watch?v=CxKXJWf-WMg
When Microsoft got into hot water during the browser war with Netscape, it was about abusing their position in one market (computer OS) to gain a position in another (web browser and web server). This by bundling Internet Explorer for free with Windows, that in turn had special binding to their IIS web server.
So when EU is looking into Google, they are looking into how their position in one market gets used to affect other markets (like say when they re-jig their search algorithms).
Hah! Good one. The best datacenter is one you never see, and never have to see.
However it does make me wonder if Schmidt knows less about software than I thought he might. The random scattered mentions of machine learning present the same image. Then again, I assume this is a person taking notes which are snippets of conversation potentially out of context.
He has a deep understanding of software. I've met him, he asked about my project, and his level of detailed questions showed he understood a great deal about software.
Maybe i am being snarky, but that seem like wishful thinking.
This may have been a transcription error, but as written I can't make sense of it. Can anyone else?
Many of them release a fair amount of their code as open source, but these sorts of capabilities are generally dependent on having absurd amounts of data, which is prohibitively challenging for anybody who doesn't run a website that sees massive traffic.
I was imagining some bit of AI that had figured out the best way to download asynchronously. Or the best way to layout a UI with code. You could maybe describe a process using natural language & it would just create a best attempt for you. You could then modify or steer it.
Seems doable, but I don't see how you'd need big data from massive numbers of users. And seems very interesting given that APIS and what they can do is much more finite than the physical world.
I went back and re-read the interview and I believe I'm channeling Eric properly.
What does he mean by this? Social people that make a good environment? Or generalist that know a bit of everything?
> You don’t hire generic people — you hire people who have had stress and achievement.
Isn't this bias inducing towards good storytellers?
> Once we decided to review all the tactics, we put in a scoring system. Sergey said, the problem is, these scores are biased.
How does google evaluate performance?
http://www.amazon.com/Pedigree-How-Elite-Students-Jobs/dp/06...
BTW, thanks for this one [1]. I'll be reading it, and similar material, really soon (had it bookmarked here.)
Regarding my comment, I should have explained myself better in any case.
I suspect all software developers have experienced some amounts of stress in getting things done on time. My problem is that I don't keep a track record of it.
Just a few days ago on a phone interview they asked me of any interesting problem I had solved. My honest non-impressive answer was "nothing in particular, I just go at it and keep going".
I have faced countless problems and when I do: I fix them. I don't write them on my diary. But now I know I should have.
One thing I do keep track are my programming and linux notes because I've dealt with so many different technologies.
And will be moving those to github and blogs since nowadays you need to be a social coder.
[1] http://kelukelu.me/interview/
Yes, the book helps. She wrote it to expose unfairness in American society, but unfortunately some people read it as a "how to break into the elite." One section explains how storytelling works in an interview context.