« More Snowy Weather | Main | Starbucks Valencia Syrup Gets the Boot »

January 13, 2007

Microsoft's Innovation DNA

Recently I was on an internal discussion thread about Google's algorithmic interview candidate analysis. There was disagreement over how valid it would be, is it really a good idea to try to find new employees who are just like your current ones, and all that.

Lingering under this, however, I got a sense that the real concern was not this particular initiative, but the feeling that Microsoft would not be willing to try things of that sort, meaning slightly out-there ideas that might or might not work...if someone proposed such a thing, it would have to go through various levels to get approval and would wind up either being turned down, or else watered down so much that it had no effect. This relates to both our ability to make changes, and our ability to attract and keep employees. It's a concern I share: not that Microsoft is or is not this way, but it is much more important that we have the ability to move quickly, then that we move quickly all the time. And certainly the company has been light on its feet in recent times (for example the change in how reviews were done, whether you like it or not, was done very quickly and broadly; flex time, as in working fewer days/from home/etc, is also catching on pretty quickly, certainly in individual parts of the company that are as large as Google; and there's the Workplace Advantage stuff).

I also get a bit of Not Invented Here attitude at Microsoft. For example, one topic that comes up occasionally is the Google 20% personal project allotment, and should we do something like that at Microsoft. But the feeling seems to be that if we do it, we can't just adopt what Google is doing wholesale; we need to improve it and Microsoft-ify it in some way (which is an admirable initial goal, but it doesn't mean if we can't improve it we can't just adopt it wholesale). For example I have seen discussion of slicing the 20% in one big chunk (e.g. 4 months on a 2-year project spent on whatever you want, then the rest on shipping the product) and slicing the 20% by people (a small group works on whatever they want all the time, the rest works on the product all the time), but not much serious consideration of the "one day a week" model that Google uses--which to me at least, seems preferable to the other two ways. Sometimes other people do something a certain way because they weren't clever enough to think of a better way, but sometimes it's because it's the best way.

Posted by AdamBa at January 13, 2007 10:10 AM

Trackback Pings

TrackBack URL for this entry:


Our team tried the 1 day/wk version for a few months, but that eventually broke down amongst the ever-present meetings and "we're behind on this" requests. I'd like to do it again, and two things we did make are (likely) making it into products. I suspect the occasional, semi-experimental/semi-clandestine attempts at this are all we'll see. I should also note that there was an (understandable) element of jealousy, since not every team in the org was doing it.

Posted by: DonD at January 13, 2007 12:47 PM

I don't think Google's is a formal "one day a week", I was really talking about the notion that it's mixed in with your regular job as you see fit (ignoring the jokes about Google people working 80 hours on their regular job and then 20 more on their personal projects).

And yes, if some people are doing it and others not, it will cause problems -- you really need management to buy in and schedule their whole team for only 80% of the available time.

- adam

Posted by: Adam Barr at January 13, 2007 04:42 PM

Random Google recruiting factoid: they just hired the former director of North America Recruiting at Starbucks to head staffing for Google's Global Online Sales and Operations Group.

From the Starbucks Gossip blog: http://starbucksgossip.typepad.com/_/2007/01/why_starbucks_r.html. This comment is mostly an excuse to mention the existence of that blog.

- adam

Posted by: Adam Barr at January 13, 2007 09:26 PM

One of the things that the Freakonomics book taught me is that figuring out cause and effect is a lot harder than it seems, even when you think you are being guided solely by data. Malcolm Gladwell found that out when Steven Levitt proved that his clean subways/broken window theory was in fact just another effect rather than the underlying cause in the decrease in crime rate.

With 100,000 applicants a year Google has to obviously use some sort of automatic screening process but I suspect that they'll find correlation does not imply causation.

Posted by: Andrew at January 16, 2007 11:08 AM