Peter Norvig at Startup School 08

Peter Norvig put it best: "The semantic web is the future of the web, and always will be." (For what it's worth, the startup school video that quote comes from is worth watching: ▶ this video )

link - (Steve Blank / Eric Ries - part 1) link - (Steve Blank / Eric Ries - part 2) link - (Steve Blank / Eric Ries - part 3) link - (Steve Blank at the Startup Lessons Learned conference) link - (Steve Blank: The Democratization of Entrepreneurship) link - (Alexander Osterwalder speaking @ Google) link - (Eric Ries speaking @ Google) link - (pg at Startup School '08) ▶ this video - (Peter Norvig at Startup School '08)

Programming bubble will pop. Automated Machine learning is taking over. Even at google, Jeff Dean has recently stated that his Perceptual AI project is aimed at reducing feature engineering in their machine learning projects. In plain terms this means eliminating the need for programming in their search engines and other machine learning focused projects like google now and google goggles (the mobile app, not google glass). The only programmers left will be to manage cloud computing infrastructure and front end programmers. As things like siri and google now take off, front end programmers will also be in less demand. Microsoft, Apple, and Amazon are also increasingly focusing on machine learning. Automated analysis of text will take over the role of sql databases as demonstrated in IBM watson. In manufacturing, the role of programming also diminishing as vision guided robots like Baxter by Rodney Brooks's team take off. All this is a good thing though, programming is slow and data is agile, as talked about by Peter Norvig in his startup school 2008 speech ▶ this video Startups should aim at doing as little programming as possible and leaving it to data crunching.

Norvig's talk at Startup School '08 went into more details on this: ▶ this video They do use SIFT (or at least a variant thereof) for finding and describing interest points, but by itself, there is no geometric matching in SIFT. There are various competing approaches on how do it, although in many cases, you can get very good results even without it. (It's very slow to do geometric matching so people often skip that step, or only apply it to the best matches.) Landmark detection is a recent "hot topic" in computer vision, and given a large enough dataset, it essentially works now for the most part.

This line of thought follows previous posts on the same subject. It's very Norvig-esq ( ▶ this video ), but there's also link and also Chris Anderson and Wired's flame bait link (that month's wired was dedicated to this subject) And like someone at the previous discussions has said, this is the base of the scientific method, not it's death