I got a nice blue box in the mail:
It seems Seth Godin persuaded Google to donate one of their new Google Mini appliances to power Joel on Software. Thanks, Google! If you use the search box, above, you can try it out.
At first I was worried that since it didn’t really use the Google infrastructure for searching, it wouldn’t be able to rank things very well. But somehow, it does. For example, if you just search for “Joel” the first result is The Joel Test, which is probably the #1 article on this site ranked by incoming links. So I guess it is communicating with Google to get PageRank information for public pages.
Overall, setting it up was reasonable: an hour or two to get the appliance racked up, most of which was spent following an infuriating procedure to get the IP address on the thing set up, and an hour or two getting the thing configured, most of which was spent following a very non-appliance-like, infuriating procedure to upgrade the software to the latest version (hey, Google, it’s on the network, why can’t it upgrade itself?)
Anyway, the point is, except for a couple of rough edges, everything was very easy to get set up, and now we have nice search results and I have a little bit more control over the results. I can do simple things like block the printer friendly versions from coming up as duplicates, and all kinds of complex things which I haven’t even figured out like synonyms, XSLT templates, and all kinds of complicated stuff. I can’t quite figure out how to make my changes to the main settings take effect right away. I tried to make it so that when you search for “mindless complexity” you get the search results for “.NET”, because I am the KING of witty, but for some reason it’s not happening … maybe I have to wait for another crawl. And the administration UI is, well, not pretty or GoogleFriendly. It’s more High Geek:
What do you think about the idea of appliances instead of software?
Correction 8/1/06: Google informs me that the Mini never phones home, even to get public PageRank information; the search results it produces are entirely based on whatever documents you told it to crawl.
