What's going on with Google robotics? | Hacker News Reader

What's going on with Google robotics?

businessinsider.com

94 pointse15ctr0n54 comments

I've seen this movie before.

Andy is a visionary.

He left Apple to found Danger which was one of the "genesis" products that led to the smartphone era. (1)

He made the next huge leap with Android and almost failed until Google showed up to help him fully execute his vision under their stewardship.

After Android became a very mature business in its own right what to do with the visionary founder?

The article states that "Page is interested in robots" - can you imagine if Larry Page came to you and said "Hey how about you literally build robots all day. You could be Tony Stark and here's $100m to get you started. We'll change the world!"

It's a hard to turn down offer, Rubin accepted and tried to recapture the magic pursuing somebody else's vision.

But it's damn tough to be a founder / visionary under somebody else's thumb, especially when you're set for life financially.

That's a story that never works out, but is played out again and again in technical acquisitions as big organizations attempt to find a place for founders.

(1) https://en.wikipedia.org/wiki/Danger_(company)

(2) https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitio...

Moral of the story - there is no great need to keep relying on the same visionary. I like Regina Dugan's ATAP (DARPA) model of kicking the visionary out after 2 years.

I'm pretty sure Andy wouldn't consider robotics "somebody else's vision."

Innovation seems to work best under threat of poverty. Because it tends to push founders to build things with a least some near-term practicality that is of economic value to some party.

If you have $100m in the bank then the only primal motivation for a goal "of creating consumer robot technology by 2020" is articles like this one by BusinessInsider questioning what value has been created.

The arbitrary deadline set at the beginning (2020) could be another motivation, but that could probably be pushed back and the project refinanced because hey innovation is hard.

How do you explain the Apple phenomenon in that case? They keep cranking out new product categories in spite of having ample money in the bank.

Apple is not a useful datapoint for anything. It's a company storied and large enough that you can't really learn anything from it.

Nobody else is like Apple, except Apple.

I dont agree that there is nothing others can learn from Apple. At one point in time, Sony stood apart much in the same way that Apple does now.

You can build teams that buck the trend, but it requires strict discipline and high-up leadership that's willing to cannibalize existing profit.

Apple doesn't hand new product categories 100$ million, they make them prototype repeatedly until they have something worthy of investing full resources in.

I'd say "promise of future status and material rewards" rather than "threat of poverty". The threat of poverty strongly induces very risk adverse behavior. One of the necessary conditions for a country to produce a silicon valley is that they have sufficient material wealth that poverty isn't a risk.

But yes, once you have the money and the fame, you probably won't be as driven.

I can't comment on how it works for companies as a whole, but it has been shown that individuals are more strongly motivated by the desire to avoid a loss than the desire to gain something.

https://en.wikipedia.org/wiki/Loss_aversion

I'm familiar with loss aversion, and it's really not relevant here. (Maybe it would be if you could fall in status by not innovating, etc.)

You can mitigate risk aversion with a starting in a niche market with a lot of money and demand or in a healthy market of seed capital and incubators.

Even in that context the threat of poverty still hangs as you're functioning with finite resources which are self-propagating.

It's usually a positive thing if the team can finance their own early growth - with a) a product that generates revenue or b) enough growth to appease investors without cannibalizing the teams creative freedom. Something which provides a runway and capacity to reinvest in a larger bolder vision which might take 2-4yrs.

The ROI of being self-sustainable is the founders can keep the team small, independent, and driven by their larger vision while still keeping them grounded in reality.

Early stage acquisitions have a similar downside as starting within Google, as they have to pursue their larger vision after the massive distraction of being acquired and integrated into a larger firm, which equally kills the necessary culture as pure poverty would.

Same with VC, purely gambling investors money on a long term vision usually leads to an early acquisition or late-game hail-mary focus on revenue - but unlike a small startup they will have much higher revenue requirements, which sideline any big visions

I think Elon Musk is an example of the opposite -- he actually put himself in threat of poverty because he wanted to innovate.

Do you have a specific source for "threat of poverty in order to innovate", or are you just referring to his large investments in his companies?

“About four months ago, I ran out of cash,” Elon Musk, 2010

http://dealbook.nytimes.com/2010/06/22/sorkin-elon-musk-of-p...

There are many other sources, and later interviews where he discusses this time of his life.

Interesting, but this is a period where he merely had a liquidity problem, not a lack of assets.

At the time, those were very questionable assets. We don't remember how close both Tesla and SpaceX came to failing. Had those not worked out, Elon really would have lost all his money.

Let's not kid ourselves. He was not actually under threat of poverty, just under the threat of losing all his assets. Worst case, he loses his cash and gasp goes to work for someone else, probably for a really high salary (maybe one of the VCs he made rich when he sold PayPal).

I think this is a little reductive. Andy Rubin left Google to form a startup incubator that's focused on hardware and robotics (playground.global). He definitely has interest in it.

I researched playground before writing my comment and stand by it.

their current investments are in wifi hardware, social connectivity, and brain-based neural networks. (1)

Going from buying 6 very specific hardware robotics companies to a diverse base hardware incubator raising money with LPs is a different flavor than being "the" robotics guy which was clearly Larrys vision for him

(1) https://www.crunchbase.com/organization/playground-global/in...

One counter example to this is John Carmack which seems to still be flourishing after leaving his companies id Software and Armadillo Aerospace, and now working for Oculus/Facebook. Exceptional entrepreneurs and visionaries by definition are exceptional.

I can't quite get what you are saying. Do you imply that Andy was more interested in other things than in autonomous robots and so he left? Or that he didn't like Page as his boss? Or that he grew complacent with all this money in the bank and decided to relax?

I think there's something to this, but I would put it in terms of Ben Horowitz's blog post "Ones and Twos":

http://www.bhorowitz.com/ones_and_twos

Larry Page is a One, Andy Rubin is a One, Eric Schmidt was a Two, and Sundar Pichai is a Two. I think Jonathan Rosenberg is a Functional One for product (he left when Larry became CEO, before I had a good sense of Google internal politics, and seems to have come back now that Sundar is CEO of Google). Marissa Mayer is a One. Make of that what you will - IMHO it explains much of the executive changeover in a couple different Silicon Valley tech companies over the last couple years.

Google's robot effort is scattered. They have two good humanoid projects, Schaft, which is a spinoff of Tokyo University, and Boston Dynamics, which is a spinoff of MIT. They're still located near Tokyo and Boston; they haven't been brought together. BD probably has better control and Schaft has better actuators for human-sized robots, and I'd expected a new machine with the best of the two technologies by now. Not happening.

If they try to move those groups to Mountain View, they'll probably lose most of the people, especially on the BD side, where the team is older. Raibert, the Boston Dynamics CEO, is at retirement age. Boston Dynamics will need new leadership soon.

Google could put Anthony Levandowski in charge. He's been with the Google self-driving car effort, and he's an original thinker. He's the one who built a self-driving motorcycle in 2003.

They need an intermediate goal, a minimum viable product, if you will. Outside of industrial robots and vacuum cleaners, no autonomous robotic product makes money. Except for self-driving cars, none look likely to do so in the near future. The most promising near term applications are military, and Google doesn't want to do those.

Yeah, there is something to be said about putting all the right people under one roof.

>> They need an intermediate goal, a minimum viable product,

My guess is that they are very careful in causing large jobs disruptions(or appearing as the cause) - without offering huge benefits to consumers, since they don't want to scare people from google.com - so they stay away from industrial/B2B robots. So this leaves them only consumer robots.

Part of the motivation behind Alphabet, perhaps?

Seeing the work of these brilliant roboticists disappear inside google without even a consumer product to show is sad. But the thought of them building robots for military is just scary.

As opposed to languishing inside a small company that is strained for resources?

At the risk of derailing the post, I have a relevant question for the Hacker News hivemind: how would someone like me get this job? (i.e. Andy Rubin's former position)

I've got a Ph.D. in autonomous robotics from the best place in Australia to have earned one. I started a consumer robotics company that exploded rapidly in a cloud of useful lessons learned. I then moved into product development consulting in order to understand the full product development lifecycle. And I'm now also the CTO of a small startup building an OS for if-this-then-that (and more) control of hardware and software devices, initially targeting Escape Room developers.

On the other hand, I'm only 31, and given how long I spent in academia, my paid employment experience is only about five years, including a 20-month post-doc.

My experience isn't all that atypical. I've met plenty of equally ambitious and credentialled people, most of them entrepreneurs too. Imagining that this sort of role is the ideal end-game for people like me, what should we do now/next/later?

I spend a lot of time looking at the history of people I admire and might want to emulate, and in almost all cases there seems to be a launching point where they make an enormous leap into an amazing role. I can't help but assume someone noticed them and plucked them out of obscurity (e.g. I'd not heard of Danger, or even Android before it was acquired) and into a high-level role. So the big question then - how to get noticed? Paraphrasing Bill Watterson, how do I find the right place to hang-out so that I'm already there when the right time comes along?

"I skate to where the puck is going to be, not where it has been."

Wayne Gretzky

Go build a consumer robotics startup that takes off and embarrasses Google, and then they will either buy you and put you in charge of their robotics efforts, or exit the market. Either way, you win.

Big companies respond better to threats than to talent. Make 'em irrelevant and they'll want you; if you try to make them want you, you'll just make yourself irrelevant.

Well said. Thanks!

Are you kidding?

Move to the U.S.

I think Business Insider misses the point. There will be many robots on the market in the coming times. Diversifying your portfolio does make sense. Also, Google does have focus in the sense that they make consumer products.

If I would work there, I'd create several robots for consumers:

* robots that can carry stuff, and yes, that means also staircases (one of my interns is doing that right now)

* robots that can clean floors properly, vacuuming, scrubbing, but also cleaning themselves! (we're doing this professionally)

* robots that can clean windows, might be quadcopters (we're not that far, we're just spraying paint for now)

Regretfully, robots are difficult to sell, so that's why I'm currently creating a revenue stream around indoor localization (which is what we solve all the time for autonomous robots). However, as Google they have plenty of time (5 years) and resources, so they can built awesome stuff.

Business Insider is just impatient.

Man I feel pretty bad for Boston Dynamics. They went from having a bright and glowing future with Government/Military contracts to being dimmed down by Google. They really seem like a company that shouldn't be tamed.

> a company that shouldn't be tamed

I don't know -- a next-gen, armed BigDog is kind of scary.

It is hugely disappointing to see what, likely, is them getting run into the ground. It almost seems self-serving that Rubin would buy all these companies, who would have been competitors for his new incubator, shackle them to Google and then leave them in the lurch to go do his own thing.

This is also what I think is going on here. Google wealthy enough to buy robot companies that might compete with its internal offerings.

The same way Google looked at internal OpenMoko efforts and decided to just buy Android. Now the problem is Android is tied to company that gets all of its revenue through advertising. I just saw an article about how typical Android apps query location thousands of times a day to sell to advertisers. Google will never implement an Apple-like "Enable location" prompt. This is their bread and butter.

This is also why I think a Google robot will fail. It'll be too tied to advertising revenues and other privacy affecting services. People will be wary about the "spying" robot in their home and businesses and will opt for a competitors product.

Google has added an "Enable location" prompt in Android 6 (aka Marshmallow).

I have Marshmallow on my 6P and have never seen it. Is this on by default for end users? Or so well hidden option that guarantees no one will ever use it?

Meanwhile, Honda[1] and SRI [2] are catching up. Boston Dynamics' Atlas is too much like Big Dog; it weighs 700 pounds and uses hydraulics. Schaft is all-electric; they use liquid cooling so they can run the motors really hard if they have to. (That's how Tesla does it, too.) Boston Dynamics is working on better hydraulics, but electric motors are going to win this one for anything smaller than a pony.

[1] http://spectrum.ieee.org/automaton/robotics/humanoids/honda-... [2] http://spectrum.ieee.org/automaton/robotics/humanoids/durus-...

> "The technology pieces we have are incredible," says one member of Google's robots team. "We just have to commit to a particular direction to go in and focus."

A more positive framing of this would be that they're doing development bottom-up: develop good pieces first, then figure out what to assemble them into later. Given that Google hasn't been in robotics very long, and that their goal is to launch a consumer robot "before 2020", this strategy seems reasonable.

The same thing that has plagued Google for years: institutional ADD.

Unfortunately, it's an extremely fine line between trying to avoid such a thing and crushing creativity entirely.

This is under the assumption that fostering that type of creativity internally is a valuable model in the first place.

It's possible they would be better off financing a VC/acquisition arm than hiring another 5k employees to be creative 'intrapeneuers'. Or some other model not yet popularized (such as the Valve decentralized approach or some shared ownership scheme).

It takes money to do things in robotics. It took $125 million in DARPA funding to get to Big Dog. It took entire CS departments at big name schools with tech support from auto companies (Stanford's VW was made computer-drivable for them by VW) to win the DARPA Grand Challenge in 2005.

You can't do this on YCombinator-scale peanuts. That's what academic robotics looked like for decades, and progress was very, very slow.

> as it tries to meet a goal of creating consumer robot technology by 2020

Wouldn't a self-driving car qualify as a "robot"?

I'm confused. Why would anyone expect that an advertising company would actually have a clue what they want to do with robotics? I think the main idea with acquisitions in robotics space was to keep Andy around, and to create yet another PR bubble, much like with Glass or autonomous cars. But then Andy left and it turned out that shit's hard. Whoop dee do. Time to switch to some other distraction.

Crazy! It would be like a PayPal founder believing he'd have a clue about how to build an electric car, or a rocket ship.

No, it would be like expecting PayPal to build a spaceship or an electric car.

I see -- it would be like if the same parent company thought it could successfully run, say, GEICO, Dairy Queen, and Fruit of the Loom.

You're reaching, but I'll oblige. It'd be like Berkshire Hathaway told Dairy Queen or GEICO to build a commercially viable robot.

Google basically runs two things profitably: Ads and Android (the latter in part because of ads). Neither of which makes anything hardware related. All phones are made by other companies, and when they tried to run one, they were losing $700M/yr.

Come on. That's like saying Apple only runs one thing profitably, consumer electronics. YouTube is wildly different from Google Fiber which is nothing like the Play store, which is nothing like web search.

Reply on news.ycombinator.com