[socal-piggies] Yahoo and Computer Science

[Steve Wedig]

http://www.techcrunch.com/2006/11/18/yahoos-brad-garlinghouse-makes-his-power-move/

Here's an interesting internal memo from an svp at yahoo talking about
problems and solutions for the company.

I've become interested in the parallels between managing
organizational complexity and managing software complexity.
Programming is the first engineering field where we aren't limited by
physical constraints. Instead we are limited by our ability to manage
complexity. So we've become really good at managing complexity through
abstraction, etc.

Redundancy breeds complexity (in any system). So we have DRY, we don't
copy and paste code, we normalize databases, and do dynamic
programming / memorization / caching, etc. Yahoo is suffering from
lots of organizational redundancy. This is on the employee level and
on the group level: Myweb vs. del.icio.us, and so forth. They haven't
refactored after their acquisitions.

We also know the importance of clear interfaces. Leaky abstractions
increase complexity, long functions are bad, etc. The organizational
parallel to this is having clear responsibilities and holding people
accountable. Apparently yahoo is having trouble with that as well.
This is partly because yahoo is a matrixed organization, which means
everyone has a product boss (ex: mail) and a functional boss (ex:
engineering). Basically each node is a member in two trees. These
things are related to the natural limitations of trees and
categorization, as discussed by lots of people…
http://c2.com/cgi/wiki?LimitsOfHierarchies
http://www.shirky.com/writings/ontology_overrated.html

Anyway, I've been thinking that a computer scientist's experience at
handling the world's complexity can be applied in almost any problem
domain.

- Steve