Charles Ferguson writes a lengthy and clearly considered piece on Google for Tech Review, focusing on the Microsoft angle and concluding that the only way Google can truly “win” is by controlling a new architecture of computing through the time honored approach of proprietary APIs. Ferguson argues that the search wars are about to enter a major battle for control of standards which simplify the increasingly heterogeneous world of search, and in such a battle, Microsoft is far better suited.
I enjoyed reading this piece, and I am sure I will read it again and again, to more fully consider its argument. But I find myself disagreeing with the premise – why, in this world of the web, do we need to be bound by this winner takes all approach to the world? It works in a resource constrained world of homogenous PCs – once a consumer has purchased his Windows box, he’s not going to easily purchase an emerging competitor – but somehow, it really doesnt’ strike me as the right metaphor for a Web 2.0 world. I do agree that Google would be well served to make its service more of a platform, and that APIs are the way to go. But I’d really be interested in what Tim O’Reilly has to say about this piece, or Tim Bray, or any number of other folks. I’ll keep my eye out…meanwhile, do read the piece. It’s a worthy provocation.
Other POVs on this piece: TechDirt, Linden, SEW, Silicon Beat
I think the real point is not that we need that sort of architecture, its that its inevitable, and whoever creates it will have market dominance. Netscape was right when it assumed the market didn’t need proprietary standards. It was wrong when it assumed that Microsoft wasn’t going to create them and wipe out Netscape. Google is wrong if it assumes Microsoft won’t try the same tactic.
Talk of APIs is interesting, but I would focus more on the architectural evolution that needs to take place, and the opportunities this evolution provides.
For instance, enterprise search is Google’s weakest link, and possibly MS’s strongest opportunity, due to its command of the desktop and workgroup.
At the web level, the architecture has to decentralize somehow. For instance, I could write a genome search engine right now, but I don’t have the resources to crawl the entire web looking for FASTA files. I need something like a subscription to a comprehensive web crawl. Likewise, others need data, or indexing, or other pieces to get their ideas to work. There’s a big opportunity to standardize and publish the different layers of search services.
I find something unappealing about a standard API for search.
This isn’t like defining a standard API for accessing webservers or databases – this API is effectively trying to define the interface between people and machines and I dont believe we’ve got that one worked out just yet.
I’d hate to see innovation in this area stunted by widespread adoption of the lowest-common denominator approach of “user provides a string of two or three words”, “server responds with first results of ‘n’ thousand”.
Can anyone access the article – appears to have been down all day today…
I thought the following sentence from Mr. Ferguson’s article was interesting:
“In late 2004, however, Google announced APIs for its advertising systems and for the Google Deskbar.”
I found references to the deskbar API, but I can’t find any mention of an AdWords API on Google. This has been a much-requested item by users of the google’s current web api.
Anyone have a link?
Thanks,
Chris
I think Ferguson is wrong on a number of points. He writes solely from the Microsoft perspective and demonstrates little acknowledgement or understanding of how Google grew so fast over the past few years.
I wrote a long post on it’s unlikely that MSN and Yahoo! are going to win over Google’s customers. Google and the Great Mousetrap Fallacy.
I think Microsoft can win almost any battle.