Wow, they are doing letters? Really? Letters? Hey is Eric Schmidt still in the building somewhere? Ask him how well Planets worked out for Sun Microsystems.
Interesting strategy, hard to second guess from the outside of course. Sun's motivation was to figure out whether the other parts of the company could stand on their own[1], it also makes it less fiscally complicated to discharge an entire group into the void. Think HP selling off the Agilent half of itself.
Generally though this sort of move is a way of containing and then "fixing" cost problems. Divestiture is so much easier once you've created the framework of a whole organization around each chunk. It can also be weirdly inefficient, at Sun each of the "planets" paid in a sum of money to IT (Bill Raduchel's organization) for "Corporate IT support" except that Corporate IT didn't work for them, they were just the only vendor you could use to get your IT services, so what you ended up with was really crappy IT work that you couldn't shop around for. It was maddening. But the 'collection of companies' design pattern requires either that you have your "service providers" that everyone uses (HR, IT, Legal) which gives little incentive for quality service, or everyone gets their own version which means a lot of excess overhead and duplicated work.
I could think of at least two other ways Google could have re-organized without bringing that pain upon them, and as Eric lived through it at Sun as well I'm sure he has an opinion.
Oh, and having one of the sub-companies get the world's #3 brand? I wonder how that works out.
[1] Answer "No" for SunSoft, "Yes" for Sun Hardware, "No" for Sun Labs.
Ha -- the Sun planets experience was the first thing that occurred to me as well. In my experience, the planets model was a total, unequivocal failure at Sun -- and the answer to the question of whether those companies could stand on their own was actually "no" for all of them (SME, SMCC and SunSoft all needed one another; the value came from what they delivered together). And Sun customers should blame the planet model for one of the worst decisions in the history of a company that had plenty of contenders for the distinction: unbundling the compilers from the operating system and separately monetizing them. But at least SunSoft, SMCC and SME came to realize that they had shared interests -- in stark contrast to JavaSoft which actually thought it was a separate company (never mind the unending welfare checks from SMI). Sun was a company prone to tribal warfare; as it turns out, dividing the company into tribal homelands did nothing for national unity.
In terms of what's happening here with Google/Alphabet, I imagine that there is some desire to see if some of these companies can make it independently while still being able to make transfer payments to support those that can't. The problem, of course, is that this builds much more explicit resentment from those that are profit-centers towards those that are cost-centers. It will be interesting to see how this works out, but hard to see how this is a stable state; it seems that shareholders will demand that businesses be spun out entirely -- or that the whole thing will have the fate of Google+ and be quietly folded up three years from now...
> this builds much more explicit resentment from those that are profit-centers towards those that are cost-centers
An additional bit of trouble, coming from the way Google's current profits are made, is that the profit centers are mainly perceived as less glamorous than the cost centers, which seems likely to only add to the resentment. It becomes much more explicit then, that the people doing the various bits of gruntwork needed to keep AdSense working (hugely profitable, but full of jobs like "sales" and "click-fraud arms race") are funding the people getting on the news for drones (not profitable, but cool). Not impossible to manage, and I'm sure they've thought about it, but does seem challenging.
That was definitely true at Sun as well. At one point, Arthur van Hoff told me with a straight face that "every E10K Sun sold was because of Java." This was in early 1998 -- and it was categorically (and demonstrably!) false. We at Sun who had the filthy task of actually making money were looked down upon by those who were responsible for spending it -- and represented one of the ultimate failures of the planet model, in my opinion.
I can't see why anyone would object to that kind of arrangement. The whole reason we do the boring stuff is because it pays for the exciting stuff. If exciting stuff could pay for itself, nobody would ever do anything dull.
If the dull stuff compensated for the lack of prestige with cash (for example), that might be true. But in reality, the low prestige tends to go with low pay, low promotion opportunities, and low recognition.
and that's why you need open allocation in a huge company. keeps everyone honest and allows economy of work to balance effort and pay.
if you don't balance people not wanting to do the low grunt work, you'll only get inexperienced or low rank employers to do it which is extremely hurtful in the medium run (say not in a year, but you'll definitely notice within a decade)
a) Is this about firewalling disappointments and legal problems from Google's core business from the new things that will drive it for the next 20 years?
b) Will they spin out Google's Europe operations in to a separate company so EU regulatory actions have minimal impact on the rest of thier business?
c) Are there other changes they are worried about to the ad business? Proposals to change the tax status of advertising deductions (US) would hit their bottom line hard, even if it if their ad auctions algorithms continue to behave as shill bids.
Will be interesting to see how this changes the main Google business, for better or worse.
To be clear I think all of that resentment is already in place and this is an attempt to bring it into the light and better manage it. I remember hearing Amazon folks thought AWS had nothing to do with Amazon and was a huge capital suck and they hated it. Now of course it's probably 30 - 50% of their valuation. Thats one way to get the "core product" folks to warm up to experimentation.
> And Sun customers should blame the planet model for one of the worst decisions in the history of a company that had plenty of contenders for the distinction: unbundling the compilers from the operating system and separately monetizing them.
Hear, hear.
The only thing Sun did that is the running for being even dumber is blowing Solaris x86. They released a pretty (for the time) awesome Solaris x86 product back in 1993, but never really got behind it. I had a Solaris 2.5.1 x86 workstation at work in 1997, and even then, my then-boss Scott Swanson was saying how Sun had blown it in terms of taking over the x86 server market.
I mean, even into the year 2006, Red Hat was telling me if their kernel crashed on a RHEL box, I had to set things up so the core file would go out over the network, instead of somewhere on a local disk. Sun had solved problems like that long, long before.
Aaaaaaaggh. Solaris x86. I know of at least one firm with 30-40,000 instances of RHEL that had been a Solaris shop. The firm tried Solaris x86, Sun essentially abandoned the platform and the firm's internal engineering board said "Never again." I'm sure that story played out again and again. What a colossal mistake.
Ugh, yes. For those who don't necessarily have the context: in January 2002, Solaris management elected to "defer" shipments of Solaris 9 on x86.[1] Technically, it was not EOL'd (and support was never removed from the operating system) -- it was merely "deferred." Of course, everyone (rightly) inferred this to be the death of Solaris x86. We in Solaris engineering knew that this was entirely asinine (and that x86 was handily outperforming SPARC) and we continued to test x86 and assert that it function (that is, if you broke x86, it remained grounds for work to be backed out)[2]. In October 2002, thanks to the work of Solaris x86 activists outside the company and Solaris engineering inside the company, the decision was reversed[3] -- but the damage was done.
The only upside (such as it was) was that the loss of trust helped accelerate the argument internally to open source the operating system, which we finally did in 2005 -- a system that lives on today in illumos.[4][5] So in the end, Solaris x86 (like many Sun technologies) represented both the company's worst (capriciously killing it) and its best (open sourcing it, giving it eternal life). Nothing about Sun was simple!
An alternative model is Samsung (and several other Korean chaebols), where upper management decided that the individual companies would be hurt by trying to act in the interests of other Samsung companies, and should act in the market on their own - e.g. how Samsung's components (touchscreens, SoCs) are sold to Apple, their mobile division's great competitor. This kind of heads-I-win, tails-you-lose strategy is enabled by the high level of autonomy lower-level groups have.
I don't think it's quite comparable though. This is chopping off companies with essentially no synergies with the core business, which are kept in a single monolithic entity. If anything the association is often negative, with the assumption that the only reason Google owns a business (e.g. Nest, Fiber) is to hoover up people's private information.
Now, there would be insane ways of splitting the company. Say putting building and operating data centers into a separate company from the company using the data centers to run services. Or splitting Android or Chrome into a separate company that'd then be funded by selling the default search engine setting to the highest bidder. If that kind of thing happens, it'd be women and children first into the lifeboats.
Agreed. And I hope that does happen. This is the best chance for Google to clear out the air around all the privacy and transparency issues it has been tackling. However I do wonder if that indeed is the plan.
For example, Google would not want to miss on the data from its Fiber or Auto or Nest projects say. It is too valuable to not include as part of a user's profile.
If anything, I feel it can lead to just making the situation even more muddy. What happens to all the Terms and conditions? Privacy policies? Would they, or even can they, stay merged?
For one thing, they're specifically keeping Labs as part of Alphabet directly, rather than as a sub-company. That seems like recognition that experiments shouldn't have to immediately pan out or immediately become self-sustaining.
For another, they've said nothing about how they plan to handle IT and infrastructure. Though that'll certainly be interesting; they'd be foolish not to take advantage of the Google datacenter and cloud services to accelerate new projects, for instance.
The recent release of kubernetes into the public domain seems to align with this. Add to that Google's cloud services and you have infra that can be used by the independently spun off orgs.
Well, yes, but Sun was more than the sum of its parts. Alphabet is actually less than the sum of its; Google is a highly profitable ad business that's still showing nice growth. The rest is a hodge-podge of startups and fantasies that provide neither revenue nor profit and have little chance of ever doing so. Most of them don't even have products. There's no reason to think that what is being taken out of Google was helping its business in any way, yet the market right now is willing to pay big option premiums on the remote chance of some future success. So what is now Alphabet would be worth more if Google were spun out entirely to its existing shareholders, because they'd get whatever Google is plus whatever people are willing to pay for the imaginary stuff. As a whole, it's just Google less a lot of its profits. None of this was true of Sun.
Technically, that's the definition of "the sum of its parts". The others are absolutely intended to have revenue and profits. Fiber already has revenue, and the potential upside on self-driving cars is astronomical. Non-G Alphabet is not a charity. It's just not (yet) a cash cow.
On the other hand, Semco is an example where this kind of thing worked really well. Their model of empowering employees, allowing innovation and splitting the company into a large number of micro-businesses had dramatic impact and turned the company around.
Maybe the differences is that they did this to encourage diversification and innovation, not to manage it.
For some reason, people really do like to try the internal market approach. It's not only on IT, companies do that with legal, accounting, any kind of creative sector (like marketing), engineering... I'm yet to see somebody try that with top management - might be a nice experiment.
Internal markets are a really nice and well tested way to gather all the flexibility and agility of a corporate environment, while losing none of the risks, transaction costs and antagonism of a real market. It's a sure way to move your company forward!
But, rants apart, I read the article as Google creating something similar to Lucent Labs, not as separating bureaucratic divisions.
At least they didn't call themselves Googlesoft or Gapple, in an attempt to define themselves in terms of their current rival-of-the-decade like Sun did.
I think you're taking the concept behind the name a little too literally...
They're not going to collect one company per letter just so they can have an "alphabet" of 26 companies. The name just implies that they have a collection, a variety, a spectrum of companies.
The alphabet is better than planets because you can always throw in more Unicode letters, while those damn astronomers took away a perfectly good planet!
It's hard for a company like Sun to have a collection, a variety, a spectrum of companies, when busy-body astronomers feel compelled to reclassify your favorite planet as a "dwarf" out from under you.
But Alphabet has a long way to go beyond the 26 English letters, before they run out of funny diacritical Unicode letters with umlauts, accents, slashes, ligatures and dingbats.
Right, that's what I was saying. The point is obviously not to collect one company for each letter. When they have 5 companies that start with G, are you really going to look at them and say "LOL I TOLD YOU SO"? The name just means variety, not a specific naming plan.
Don't misunderstand me, it isn't dislike, its amazement. Some things really do well under multiple companies, think BASF or Siemans where the core that holds them is something common to all the sub-companies. But historically tech companies do not prosper in that model, and generally for a similar reason, there isn't anything in common between the companies except maybe there are smart and talented people working there.
The original Sun vision was that Sun Hardware would sell SPARC architecture machines running any OS, SunSoft would sell Solaris for any type of hardware, Sun Labs would license their innovative research rights to anyone who wanted to productize them. Except ...
Nobody really wanted to run their OS software on Sun hardware because SunSoft had an inside track on what was coming out. Sunsoft couldn't really make Solaris on x86 just as powerful as it was on SPARC because it would undercut Sun's HW arm, and Sun Labs? They could never really license a technology to a Sun competitor without the other two companies vetoing it. The whole reason FirstPerson was created (the shell company that was developing Oak which became Java) was to keep the politics and other constraints off what could be done with the software. And when the group told Sun it was going to ship first on "Chicago" (aka Windows 95), Scott McNealy blew a gasket.
Pretty much everyone I've talked to about that decision in depth thought it was a net negative for Sun and that had we had a chance to do it again, they would do something different. And since Eric was right in the middle of that, I would have expected him to advise Larry and Sergei that it wasn't the best choice. And maybe he did, and maybe they have a scheme that will avoid the problems Sun had. I'm surprised though that they chose that route.
The difference I see is that Google isn't unbundling their core businesses from each other; we're not getting something like Sun splitting SPARC and Solaris.
Search, Android, Chrome, YouTube, Maps, etc. are all staying under Google. The stuff that's going under other Alphabet subsidiaries is the stuff that's totally disconnected from Google's core businesses, like Calico, Life Sciences, etc.
Joel Spolsky's Strategy Letter V is something that I think is interesting in this Sun conversation
> Sun is the loose cannon of the computer industry. Unable to see past their raging fear and loathing of Microsoft, they adopt strategies based on anger rather than self-interest. Sun's two strategies are (a) make software a commodity by promoting and developing free software (Star Office, Linux, Apache, Gnome, etc), and (b) make hardware a commodity by promoting Java, with its bytecode architecture and WORA. OK, Sun, pop quiz: when the music stops, where are you going to sit down? Without proprietary advantages in hardware or software, you're going to have to take the commodity price, which barely covers the cost of cheap factories in Guadalajara, not your cushy offices in Silicon Valley.
Joel has always seen things from Microsoft's point of view. And we are actually living in the world he mocked: both hardware and software, on the server side, have been largely commoditized.
One could say Sun read the tea leaves of the industry correctly, they just didn't couldn't shed their history quickly enough to execute on it. Cloud services are the new proprietary and Sun had begun building that shortly before Oracle came knocking.
Awesome insight and backstory about one of SV's iconic companies, so thanks for sharing. Maybe the new Alphabet structure, despite similar failures at Sun and other tech companies, is the best option for maximizing innovation even though it also carries a high probability of failure (an "alpha" bet if you will, hah). Or maybe they did devise an ingenious scheme to neutralize the shortcomings of the Planets model, but it would be pretty cool if they didn't and are knowingly exposing themselves to risk and ridicule in the name of innovation.
I fully expect that they have thought it through. And I do presume that Eric shared all of the pitfalls that befell Sun when it tried Planets. So yes, no doubt the idea is fully vetted from that perspective. Given the pain for Sun, and Eric's direct experience of the same, I am surprised that this is the path they are taking.
If their goal is to maximize the odds of innovation, not minimize the odds of failure, which path would you advise? And please share more Sun stories. These were entertaining and informative. It should have been Sun leading the cloud revolution, but that's the Innovator's Dilemma for you. :(
Interesting strategy, hard to second guess from the outside of course. Sun's motivation was to figure out whether the other parts of the company could stand on their own[1], it also makes it less fiscally complicated to discharge an entire group into the void. Think HP selling off the Agilent half of itself.
Generally though this sort of move is a way of containing and then "fixing" cost problems. Divestiture is so much easier once you've created the framework of a whole organization around each chunk. It can also be weirdly inefficient, at Sun each of the "planets" paid in a sum of money to IT (Bill Raduchel's organization) for "Corporate IT support" except that Corporate IT didn't work for them, they were just the only vendor you could use to get your IT services, so what you ended up with was really crappy IT work that you couldn't shop around for. It was maddening. But the 'collection of companies' design pattern requires either that you have your "service providers" that everyone uses (HR, IT, Legal) which gives little incentive for quality service, or everyone gets their own version which means a lot of excess overhead and duplicated work.
I could think of at least two other ways Google could have re-organized without bringing that pain upon them, and as Eric lived through it at Sun as well I'm sure he has an opinion.
Oh, and having one of the sub-companies get the world's #3 brand? I wonder how that works out.
[1] Answer "No" for SunSoft, "Yes" for Sun Hardware, "No" for Sun Labs.