Loon and Wing graduate from Google/Alphabet X | Hacker News Reader

Loon and Wing graduate from Google/Alphabet X

wired.com

133 pointsjpm_sd57 comments

This (almost) exact story seems to resurface every couple of years. Given that Google has to allow (or perhaps) initiate the story and access, I wonder if there's a specific type of situation in which Google's PR team decides that it's time for another "Inside X" story.

In this particular case, it's because Loon and Wing are "graduating" (spinning out), although that information is a few paragraphs deep. And it's in the URL, but not the article title.

Ah I see. Since the title you submitted with is actually the language of the URL, we'll put it back.

Earnings season?

I wonder if it's when they have trouble hitting their recruiting numbers. There's a popular theory that X is the carrot dangled in front of both current Google employees and potential hires-- "Look at this cool stuff you could work on!"-- while in reality they usually get sent to the trenches to squeeze out 0.003% gains in revenue-per-query.

Nobody I have ever worked with at Google has had this carrot dangled under their noses. It certainly wasn't dangled under mine when I was hired. I don't dangle it under the noses of anyone I interview. I think most people with a basic understanding of arithmetic realize that the overwhelming majority of the company does not work in X, and they should not expect to, either.

I have zero interest in working in X.

The reason hiring is hard is because the interviews have a low pass rate, everyone else is also hiring, and salaries + benefits among top tech firms in an area are fairly similar.

> The reason hiring is hard is because the interviews have a low pass rate

But the pass rate is up to the company. Either the wrong people are being brought in for interviews, or Google is looking for some x-factor that nobody has.

I work for Google and conduct interviews for Google. No magic x-factor, just a good command of data structures and algorithms, plus the ability and stamina to express it in a full day of interviews, plus the luck to have a good night's sleep beforehand, not have an off day, etc.

The strongest factor in my opinion (after pure luck) is the rapport you're able to build with the interviewer. Doing that of course requires doing well in the technical questions, but the end-goal shouldn't be seen as doing well in the questions, but seeming "Googley" to the interviewer. Googleyness in this context tends to include not just whether you did well on the questions, but things like your sense of humor, ability to communicate in a positive and cheerful way, whether you seem mature and responsible, and other such little things the typical techie would like in another person. Interviewers are also human, and they can only do so much to change how they perceive another person.

Source: Have interviewed with Google a couple of times, and coached a couple of other people for the same.

I can't speak for other interviewers, but I barely take this into account. As long as someone isn't actively antisocial (has never happened to me) or unprofessional (has happened once or twice) I file feedback exclusively on technical merit and communication. There have been times where I liked someone and wanted to give them good feedback, but couldn't.

> I barely take this into account.

> I file feedback exclusively on technical merit and communication.

I appreciate your comment, but my point is that there's no way to actually do this, even if your goal is to. It may seem that you're being objective, but the objective facts can only influence the subjective opinion that's ultimately responsible for the decision, not the reverse as we like to think.

For example, if an interviewer likes the interviewee, they would actually think that the person did quite well in the technical question, and not if they don't. We first form an opinion about someone or something and then look for "objective" facts to justify it, not the other way round.

The only way to really be objective here is to use a point system and hard thresholds which completely removes the human in the loop, like SAT.

I recommend "Thinking Fast and Slow" and "Why Buddhism is True" which go into this idea in detail, and in general about how poor we are at being unbiased and rational. You'd be surprised at how deep our irrationality goes even if you already have some idea.

And that's why Google's recruiting has the extra step of a hiring committee which didn't meet the candidate. Interviewers fill out interview feedback based on how their interview went with the candidate, and there's guidelines to try to reduce bias (no gendered pronouns in feedback, the candidate's code is included as-is in the feedback, etc) and the committee makes the final decision based on the feedback.

Sure, what is included can still be slightly biased, but if you're talking about non-technical things at the beginning or end, it won't really make it into the technical interview feedback, and probably will have little to no effect on the committee's decision. At the end of the day, the problems you're asked have optimal solutions and your code will either work or it won't.

Okay, of course, I didn't mean to suggest that I'm a robot - I meant that"Googleyness" is not a conscious criterion. I'm sure charisma etc. can influence me a bit. But when I've asked the same question 50 times, and I have to write up feedback in a structured way (I'll be vague about this to play it safe with my employer, though it's nothing exciting or complicated), there's only a quite limited amount of wiggle room. I certainly wouldn't advise a candidate to optimize for rapport over whiteboard coding or algorithms. Like I said, I've interviewed candidates with whom I had good rapport - with whom I'd love to have a beer - but I still couldn't write strong feedback for them.

I don't think your point is totally off base, I just think you stated it too strongly in your previous comment.

On the flip side, a good interviewer should be building rapport with the interviewee regardless of interview performance. This is especially important for earlier interviews in the day, because you don't want the candidate to get flustered and ruin future interviews.

Even better is to use that rapport to then disguise successively large hints into the interviewing process, so the candidate can't obviously tell if they've failed horribly... again to avoid tilting them for future interviews that day. Give them room and time to explore the problem space on their own, but don't let them sit in silence making no progress for 30 minutes making it painfully obvious that something is very wrong.

A single bombed interview might not kill the whole packet, so it's best to insulate.

> but seeming "Googley" to the interviewer.

There is no such thing.

> but things like your sense of humor, ability to communicate in a positive and cheerful way, whether you seem mature and responsible, and other such little things the typical techie would like in another person.

Sounds like you're hiring actors rather than people able to do good technical work.

Google as a company seems to have none of these qualities in the aggregate so it looks to me as if your process is broken. Besides the obvious ethical challenges.

I'm not saying I approve of this. Just stating how things are. I'm not an interviewer and don't represent Google, but I doubt other big companies have any better process.

If the hiring process is able to identify a good hire whereas just the resume does not (which is a big if, but companies assume that it does), then it totally makes sense to bring in a lot of people to interviews.

The pass rate is indeed up to the company. I never said otherwise. It's a very low rate.

The answer is that google doesn't give a shit about false negatives.

They know they will get enough people applying several times that they can just toss out qualified people all day. Eventually enough get in.

Unless of course the hiring process has a systemic flaw and there's an entire class of people that are being rejected that could be game changing.

(I actually don't have an axe to grind here - it's certainly not related to any obvious grouping - I just wonder if interview and selection might be so deeply flawed that a the occasional random lottery injection might improve the overall result.)

There are effectively random lottery injections via acquisition/acquihires. Source: I was/am one.

I thought they made people re-interview?

They do but it’s at least a lighter schedule.

I also think the criteria are different. i.e. I believe when evaluating the results it’s more strongly biased to avoid false negatives unlike the regular process which is more about avoiding false positives.

Or maybe I just have impostor syndrome, who can say...

Would you say it's particularly hard to work in Google X, or just a smaller part of the company with fewer headcount requirements?

i can only wonder what kind of sweet deal the employees of the graduating project get to stay with the project and the new company as otherwise being kicked off the mothership can be expected to be a good trigger for job change.

I believe graduating from X doesn't mean they are no longer going to be under Alphabet rather that they've graduated from being a moonshot project under X.company to being their own thing, still under Google/Alphabet and all the benefits that entails.

If anything I imagine its a net positive for the team, though I'm not a Googler and don't really know the internal dynamics of the situation.

I think many employees would be very interested in their equity compensation...wonder if employees are awarded RSU's in Alphabet or the subsidiaries or ...

Alphabet

Good to see Loon graduating into its own company. Project Loon was very instrumental in connecting Puerto Rico after the hurricane [1].

[1] https://www.engadget.com/2017/10/20/project-loon-att-puerto-...

Was it? The numbers Look posted didn't seem super impressive given the number of people on the island. They also said nothing about reliability. I want to like Loon, but don't have much data to really go on.

Most of the comments in your history are Google related, or anti-apple

Are any of these moonshot projects trying to counteract human-driven climate change?

> The Foundry uses this intense interrogation to root out the things that could kill a project down the line, before X has poured in piles of money and time. Take Foghorn, X’s effort to create a carbon-neutral fuel from seawater. The tech was amazing and the problem was huge, but two years in, the team realized they had no viable way to compete with gasoline on cost—and were reliant on technology that was closer to research than development. X killed Foghorn, gave everyone on the team a bonus, and let them find new projects to push.

There's a salt battery project in X. Makani is also in X. There's a thermal pump company that graduated from X last year. There's also a team at Google working on Fusion research. Probably more I don't know too.

Thermal pump -- is that Malta?

Not sure if thermal pump is really the right term, but I meant Dandelion.

TLDR: big company tries to remain dynamic by running its own internal innovation program

Too bad it has never worked too well for any other big company, for structural reasons ("Internal incubators")

Actually it worked very well for many companies. The Amazon Kindle came out of Lab 126, Amazon's innovation program.

The idea isn't just to remain dynamic, but to allow for new products and services to be developed despite posing a threat to the company's main cash cow.

So Lab126 was completely independent from most of Amazon, including being based in the Bay Area vs. Seattle. This allowed the to develop a product that went directly against the main cash cow at the time (selling physical books) and make it to market.

Microsoft had one as well, but would always fail because the existing businesses heads held too much power, and would always kill the products. Just take a look at the history of the Surface.

> The Amazon Kindle came out of Lab 126, Amazon's innovation program.

I know this from all the com.lab126 Java packages in the firmware :)

Well, ereaders and ebooks are really obvious things to develop if you are in selling physical books business.

Not if you're Kodak and making film.

Or, at least, the technology might have been obvious (they did have a hand in developing digital cameras), but the idea that they should bet big on the new technology eventually replacing them evidently wasn't obvious.

It is even worse than that for Kodak -- Steven Sasson actually created the first digital camera using a CCD sensor while working at Kodak [1] and Kodak owned the patent for a digital camera [2].

[1]: https://www.seattlepi.com/business/article/Kodak-engineer-ha... [2]: https://patents.google.com/patent/US4131919

For a company in Kodak's position, trying to reinvent themselves may not have been the best idea. Another possibility would have been to recognize that the basis for the company was going away and attempt to milk it for as much money as possible as it died. If this meant paying large dividends then shutting down or carving up the remnants of the company, so be it.

The PARC was (contrary to the popular myth) a huge success for Xerox, as the laser printing technology invented there turned out to be a great money maker for the company.

Except when it does to the extent that the common term for this approach ("skunk works") derives from a big company that did exactly this.

you're assuming the point of R&D is soly to launch new products.

There's other benefits. Such-as institutional knowledge that can be used to determine where to invest VC money / acquisitions.

Or the strategic part where you leverage R&D spend to sign quid pro quo contracts...who's earnings might boost some quarterly earning metric.

It's worked plenty of times. Intuit is highlighted in Lean Startup for having it's internal innovation programs playing a pivotal role in the company. The CEO at the time says that it saved the company from stagnating growth problems.

do you know where I can read about this, I've always been curious about that

Try industrial organization in econ

Thanks a lot

also cisco - http://www.businessinsider.com/why-cisco-showered-three-men-...

This is wrong. The article even mentions this. Look at Bell Labs

I am not surprised that an PR article would try to spin the positive side.

There are rare exceptions in anything - but the existance of lottery winners does not mean the expected payout of a lottery ticket is positive.

Bell Labs used its monopoly profits to fund research - which was good for research, but no so much for Bell.

If we consider AT&T as Bells successor, it only allowed Bell to stay relevant. Still, AT&T now is nowhere close to Bell it is heyday. Same for Xerow now. Research doesn't pay as much as companies seems to think.

While google position in online advertising is as close to a monopoly as legally possible, it will not change the structure of the incentives. The odds are against them.

But maybe they do not aim for the kind of success they claim they will have, and only want to stay relevant, like Microsoft is now. That is more possible, even if it will be hard to pull.

Personally, I think most R&D money that companies like Microsoft (or Google) have sunk into internal R&D could have had much better uses. But it's not my money! And eventually white elephants get butchered for the meat, so no big loss there.

> Research doesn't pay as much as companies seems to think.

Some companies are constantly researching ways to improve their products. Facebook, Amazon, Google, Microsoft, and Apple come to mind.

Some companies are constantly researching for new ways to find their resources. Every oil company comes to mind.

I think your claim is way, way off.

By "research", I meant it in the context of this discussion : "internal incubator" style R&D to try to spawn a brand new line of products totally outside of the current business line.

Ex: tablets for Microsoft in 1994. Google driving cars.

Android, Chrome, Maps, YouTube, GMail...

The "current business line" of Google was Search. Full stop.

Some people thought it was absurd when Google started doing email, or smartphone OSs, or browsers...

I could go on and on.

Microsoft doing gaming consoles. Amazon doing web services rather than online storefront...

Waymo has a good shot at being huge.

Reply on news.ycombinator.com