Engineering produces wisdom

A software organization can generate competitive advantage in many ways: network effects, branding/counter-positioning, switching costs, etc. One historical way to build power has been creating leverage through engineering: an engineering organization that can generate value quickly or cheaply is better than one that can't. Gauging the efficacy of an engineering organization is, sadly, an unsolved problem, but three useful aspects are: speed stability cost These must be considered as vectors, not points; I find myself constantly amazed at the efficacy of LLMs while also being surprised at how and why they fail a given task in uninteresting and inscrutable ways. Models from frontier labs are currently very cheap (but could soon not be); they are very fast (but tend to lower their own velocity over time); they are good at improving some but not all aspects of stability. Just as Warhol said that a Coke is a Coke and you're drinking the same bottle as the president, there is little…

Commenting on Bubbles has been removed. Why? If you want to talk about this post, write to the author, use the comments on their blog if it has them, or reply with a "RE:" post on your own blog. Suggestions? feedback@bubbles.town