HP Labs as "hit factory"?

HP Labs has been working its way through a fairly major reorganization the past few months, and the PR machinery is using that to raise some visibility in the press. This Fortune article has a nice quote that sums up my feelings pretty well:

"Along the way, there are many things that could go wrong. Chuck Geschke, who left Xerox PARC with his friend John Warnock and 25 years ago co-founded Adobe Systems, notes that no one has yet cracked the unique code for making advanced research consistently pay off. “There must be something special, because not everyone’s able to do it as well as they anticipate,” Geschke says. And perhaps there is no perfect formula at all. “You have to accept the fact that if you’re really pushing the envelope, some of what you do won’t work out.”"

FORTUNE: Big Tech Turning an idea farm into a hit factory

I think Geschke pulled his punches and really should have said "because nobody ever does it as well as they hope to." But his skepticism, like mine, is clear.

Any very large company like HP has evolved countless systems to protect against risk, and there are countless employees who carry out that mission. HP is now a $100B company, and if you divvy that up that means there are theoretically >1000 (!) $100M businesses in themselves, each of which will try to defend itself from any perceived threats. If a disruptive incubation cannot steer clear of all of these, in addition to the corporate-wide functions (legal, finance, etc) that represent them all, that project is doomed. Thus the odds of success for incubations in a BigCo would appear to be tiny. In my mind this is the biggest difference between a BigCo incubation vs. a small independent startup. The BigCo may think it gives autonomy to the incubation, but in practice it never does.

I've been involved with several incubations in my years at HP, most of them in HP Labs, and the main impediment to success has always been other parts of HP, not external competitors. Unless a makeover like this includes a dramatic revamp of how the disruptive activities interface with those countless existing functions inside the BigCo, you will likely end up with the same end result: a lot of frustration all around.

Call me skeptical, but I've always wanted to make this work inside HP, and intend to keep trying :-).