Peter Norvig, director of search technology for Google gave a talk on the future of search. Much of the talk was banal discussion of the merits of having lots of documents to applying various forms of brute force search algorithms. But, he show a really interesting slide indicating that the size of the web was somewhere between 10 and 100 terabytes.
Some quick math reveals that it costs only $10k to store the bottom end of this spectrum. The hard disk and bandwidth variants of Moore’s law tell us that we are rapidly approaching a time when we will be able to download the entirety of the Internet onto people’s home computers. Once we are able to do that, Google will be competing as a commercial software provider against a hoard of open source software developers steadily improving home search technology and customizing for the needs of various types of users and content. I asked Norvig about this and he claimed that Google’s commercial software will outcompete open source. His answer is incredibly ironic given that Google itself is built on top of linux and other open source products and Google’s theme “Do no evil” is an attack on the commercial software world of Microsoft.
In any case, whether the threat is other commercial vendors or the open source world, the future of Google’s search business looks iffy at best. Admittedly we are some years away from individual home users being able to download the entire Internet, but there are lots of intermediate versions of this threat including a proliferation of other web sites that can suddenly afford the costs of downloading the entire internet and whose costs are covered by e.g. being shared by the users of a particular ISP or company. This option may be particularly attractive to businesses that would prefer not to have the searches their employees do be visible to Google Inc. or anyone packet sniffing along the way. The other major form of competition may be the download of particular subsets of the web optimized for particular markets e.g. crawling only in the finance world to answer queries like “aggregate revenue of of companies with the word ‘chemical’ in their annual reports or crawling Friend-of-a-Friend files looking for a relationship….
Either way, if you own Google stock because you believe in its future in the search business, you might want to sell. However, you should be aware that Google’s revenue base is shifting dramatically away from advertising on google owned sites to advertising on non-Google sites for which Google serves advertising (“adsense”). There is no reason not to expect this trend to continue because the network effects of such a business make it incredibly hard to dislodge. and every reason to think that Google’s work in search is becoming something more akin to basic research than it is to relevant part of an active business. So the expected revenue of the business may be protected. The only doubt on this score comes from my friend L, who, when I made this observation to him, expressed dount that Google’s techie anti-marketing corporate culture is going to prepared for this transition. The cold response I got from Norvig on this question is consistent with L’s observation. I don’t know enough about large scale corporate “change management” initiatives to know what risks this adds to the value of Google’s stock. But caveat emptor.