Astro Teller on Innovation

Google's Director of New Projects speaks to Singularity University's student teams on creating a culture of innovation, and how "common sense" is often at odds with creating this culture.

Category:

Education

Tags:

• Astro
• Teller
• Google
• upload2
• copy

[NaturalCurrentSavior]

Fantastic presentation... Thank you!

Michael Evingham

Natural Current com

949 636 4358

[bobryskamp]

Fascinating stuff. My favorite bits:

15:20-20:00 on taking lots of little bets

38:20-43:15 on saying no to 99% of new things

[rolandtritsch]

+1 - Creating an instance is "me-too". Innovation is creating a new class. (59:00)

[ellenjhunt]

And - the most corrupt banker in the region was a man from England who came from a Western NGO. He netted on the order of $17 million for himself taking kickbacks and bribes.

This sort of thing was pervasive.

[ellenjhunt]

I tend to agree with him about NGOs. I worked in Central Asia for 5 years. I saw intense corruption and great damage from NGOs.

Examples: Red Cross and Drs w/out Borders being used by insurgents as a free medical care for themselves and the locals. This includes complete survival infrastructure. Without that, the wars would have been over long ago.

Western NGO staffers teaching need for fair taxes. But NGOs have a law that exempts them from taxes while staff make 3X-10X locals good wage.

[ellenjhunt]

I can think of several instances in biotech where fantastic innovations have not gone anywhere for totally different reasons than he discusses.

He is quite wrong when he says "Dr.s aren't supposed to use it." Off-label is 60% of all pharma. Dr's are allowed to do it!

The "torturing themselves" bit is classic entrepreneurship talk. It's the "pain" thing. (Yes, I taught MBAs for 6 years.)

[ellenjhunt]

It's interesting to note that "Me too!" is the REASON why venture capital doesn't make far more money than it does. Because venture capitalists are lemmings who jump into the "hot sector" they GUARANTEE that most bets in the sector will fail because every sector has limits.

[ellenjhunt]

But, he is basically right. Years ago my team was the first in the world to make a full-scale factory automation system work. A team of 4 math PhDs told us that they had spent 10 years working on the body-bank remix algorithm and had proved it was impossible. We solved it in 6 months by changing the rules. We simply backed up the algorithm to include initial scheduling of the automobile to be built in the first place.

[ellenjhunt]

He has some good ideas. But there are holes. For example, he says that the balancing between bonus for killing your own project and keeping it is emotional investment. This is false. The balance is the present value of the bonus versus the risk assessment of the probability of having a job in the future.

[ellenjhunt]

He has the story about DOS and IBM quite wrong. Microsoft got the contract to write OS/2 and convinced Cannavino they should write it in assembler. This was a deliberate strategy to damage IBM. Microsoft used the OS/2 project as the farm for engineers for their own NT project. NT was the from-scratch rewrite and it was done in C/C++. This allowed Microsoft to come out with NT and by the time IBM figured out the dirty trick, it took too many years to catch up.