Why Google Doesn’t Have a Research Lab | Hacker News Reader

Why Google Doesn’t Have a Research Lab

technologyreview.com

68 pointssherjilozair42 comments

Google doesn't need a designated "lab space" because Google is a laboratory.

It might be different if Google were a chemical engineering company, but it's not. It's a software engineering company. Anywhere they've got computer scientists, whiteboards and computers is a lab. And that's basically every building at Google.

You can not say that is always true. Microsoft is a software engineering company as well, but they do have a dedicated R&D team and offices.

Just because MS has a separated division doesn't mean Google has to follow. A lot of the google come out as both product and paper. That's already a research lab. Why do we bother to discuss this in the first place? I really don't see any difference in having a dedicated research lab and a place where research and products are mixed together, true a lot of employees at google don't publish papers and they do boring works like fixing bugs.

MS didn't start that trend. Many companies used to have research divisions, large ones even...HP, Xerox, DEC, intel, Sun, IBM, even Apple had something. That is almost gone or rapidly shrinking...

I don't mean to disrespect these companies, but they are much older than Google. It seems like from the sart Google didn't care about the division. They knew they always have some "research" going on. Maybe founders' background was the founding culture.

Microsoft also does a lot of hardware research and innovation: XBox Kinect, Surface, etc.

To some extent, hardware research requires more dedicated facilities than software research.

Who said it was always true? I certainly didn't. I made an assertion about Google, not Microsoft.

> It's a software engineering company.

This sort of implied a generalization.

Additionally, it seems most Google employees are already trained in the arts of secrecy about their day jobs. No need for extra shielding for those working in R&D.

Working at Google sounds like a Fun time, where you can explore your inner child--and learn how to make money on ads. I know all the Google employees are geniuses, but a genius without the right guidance(Scientitists--who could care less about advertising, just the scientific method) might end up just doing what Google rewards them for--making money. I applaud Google for letting us use their API's, but I feel restrictions are coming, along with fees. I'm not a huge fan of google for one reason; they ruined my Privacy. By the way, you can get off Google street view by telling them you have security concerns--at least for now. The minute DuckDuckGo gets a little better--bye to my back stabbing friend--Google. And yes those Google glasses will make some people angry(in kid's speak--you just might get a drink thrown at you, and no--even if they start to make them look like EMO Glasses, some people don't like their picture taken without their permission.)

Google is definitely moving the direction of Twitter with their APIs. They are becoming very locked down.

In the recent Android Wear release, for example, notifications have always supported a thing called RemoteViews - a way to put buttons and the like in a remote process like a widget and a notification. Android Wear intentionally strips out everything but text and icon, however. I wish we still had the old Google who would have let us have powerful APIs and said, well, users will rate up apps that do or don't do good things. Proof by data.

Instead we have another Twitter. I hope they don't start having bots analyze all the API users and kill ones due to robotic misclassification like happened to many of my Twitter API apps. Or user token limitations like killed so many more Twitter apps, geh. I guess they do do both already, though. If I open a bunch of tabs in Google Groups quickly, I get banned because they think I'm a robot. My Google Apps for business accounts got made non-free and charged for each user now...

Sadly the days of asking for permission are long gone. Every day we're recorded on dozens of CCTV camera, car dash-cams (and similar worn by bikers / cyclists), traffic cams, police severance cameras (in larger urban towns / cities) and so on. You can't walk into a single shop nor office without being recorded. you can't even talk a walk outside, nor a drive, without the chance of being recorded.

I'm not saying it's right, but that is how it is these days.

You can blur your face with LEDS if cameras are your major concern. Someone should make some stylish hats.

This is a big strategic question. Do you want researchers closer to the market, paid by the person whose P&L they could impact? Or further away, so they can think longer term?

Other companies go through this a lot. Procter & Gamble used to have centralized research. To improve time to market, they decentralized research, and had them report to business unit heads. This was great for singles, but when they wanted more home runs and cross-category innovation, they decentralized again.

One can look at Xerox PARC as the penultimate example of the dangers of isolated innovation labs. They created the technology that everyone else monetized. AT&T struggled with this somewhat too.

Xerox PARC was not dangerous. The danger for Xerox was the different between seeing an opportunity for the future or a threat to their current business model. If PARC had never existed, the same things with minor differences would have happened some years later.

The danger was not the invention it was wasting money without realizing the benefits.

Exactly. They invested in a future that other realized instead of them.

Would it have happened anyways? Who knows, probably.

Who was harmed? Their shareholders, and all the employees who eventually lost their jobs.

Centralized research labs give researchers the opportunity to look at different ideas across the company and connect ideas. This is almost impossible if all researchers are siloed in decentralized research groups.

Also, there's 'why can't we have both?'. In honesty, Google, like most tech companies today follows a hybrid model. The public statements of executives are not to be taken as gospel, because their primary audience are the company's shareholders. In reality, you have PhDs in Google X working on moon-shot problems (product focused or otherwise, they are long-term projects) as well as in direct P&L divisions. Centralized research labs also have their own versions of P&L metrics (e.g. number of papers published, product $s resulting indirectly from research, number of patents, etc.). Financially, centralized research labs working on moon-shot problems are a hedge against short term fluctuations and gives companies an opportunity to pursue long-term trends. Also, in companies that have both, there is nothing to prevent researchers from moving between centralized research labs and decentralized when opportunities present themselves (e.g. product commercialization/scale-up project resulting from research). I think it is dogmatic to pursue centralized or decentarlized exclusively.

This is purely semantics: "Google’s research boss, Alfred Spector, has a small core team and no department or building to call his own."

So no department but he has a "small core team" and presumably Alred and his "small core team" have offices? This is such a non-story...just PR fluff to make Google appear unique and more "integrated".

Are we actually supposed to believe the "labs" at other tech giants are some sort of ivory tower that they never leave?

There is one, it is right above Blaze cafe (1225 Charleston Rd). There's a huge sign by the visitor's area: "Google Research & Development".

A more in-depth look from the Communications of the ACM, by Alfred Spector, Peter Norvig and Slav Petrov, "Google's Hybrid Approach to Research": http://cacm.acm.org/magazines/2012/7/151226-googles-hybrid-a...

Then why they have this: http://research.google.com/

``Research at Google is unique. Because so much of what we do hasn't been done before, the lines between research and development are often very blurred. ''

> Research at Google is unique. Because so much of what we do hasn't been done before...

How is that any different from regular research?

The article is referencing the physical space, not the existence of a research team/department.

What's Google X if not a research lab?

A playground for extremely smart people they don't want their competitors to hire.

X is still very much run by Sergey. On the more experimental projects (i.e. not Glass or self-driving cars) you have a fair amount of freedom to experiment, but if Sergey wants something to happen a certain way, it happens that way.

(For that matter, the rest of the company isn't all that different: you have a fair amount of freedom to experiment, but if Larry wants something to happen a certain way, it happens that way.)

BTW, I sent you an email about meeting Larry/Sergey a while ago. I just added rachelbythebay to the list too.

> A playground for extremely smart people

Which is pretty much a definition of a good research lab.

I believe that the second clause in hendzen's reply is the point of the statement.

> ... they don't want their competitors to hire.

i.e. that the purpose is not to get them to do research for Google, but to stop them doing research or development for anyone else.

Unless they're paying them to goof around on tennis court, I don't think it matters. It's not really important for who they're doing the research (except maybe that Google is the least evil company I can think of), because ultimately whatever they invent will benefit the whole mankind.

Sounds a lot like research to me

The article might describe it as a 'product development group' but really that's just a lab for engineers rather than scientists. It's not like Los Alamos where it was science-centric research, rather it's more like PARC or what Stark Labs would be like, in that it's 'product'-centric because it's the engineering rather than scientific version of a research lab.

According to the link, a product development group.

It's a lab.

Yes, it is, specifically for high risk high reward projects and technologies.

Off topic: Can anyone list ongoing software based research projects, that are critically interesting, preferably in the sphere of compiler design, low level system software or high performance computing?

Off topic: Rust from Mozilla is provably the first thing I think of outside of Google that matches the compiler part.

http://llvm.org/pubs/ ?

In terms of compiler design, Graal/Truffle springs to mind.

Hitting maybe all your targets: http://rosecompiler.org/

Well, I think Google should have to make his research lab..Google’s research lab, Alfred Spector, has a small core team and no department or building to call his own,according to MIT Technology Review.

Reply on news.ycombinator.com