« January 2006 | Main | March 2006 »

February 28, 2006

Retiring Jeeves

Apparently Ask.com is retiring Jeeves, their old cartoon butler:

From management's perspective, Jeeves had morphed from an endearing mascot to an exasperating albatross. That's because the butler's image conjured memories of a long-bygone era when Ask.com promised to deliver simple answers to questions posed in conversational language.

The question-and-answer approach never worked like engineers envisioned, prompting Ask.com to shift direction. The company now believes its search tools are as good, if not better, than Google's -- a message that Berkowitz believes would be difficult to convey as long as Jeeves stuck around.

"Never worked like engineers envisioned", indeed! When Ask Jeeves first appeared, I was working for a stodgy financial company back in Atlanta. One day, while trying to get to the bottom of a thorny compiler problem, I innocently asked Jeeves: "What is peephole optimization?" (*)

Jeeves came back with: "Did you really mean to ask: Where are the best places to cruise for gay men in Alabama?" I was pretty sure I hadn't meant to ask that, no.

Later that day, I mentioned this experience to my friend Colleen. Being much more clever than I am, she wondered how Jeeves would respond if you asked "Where are the best places to cruise for gay men in Alabama?". We tried it, and Jeeves came back with the same question, but also with: "Did you really mean to ask: Which cruise ships have the best sanitation?" Once we stopped laughing, we put that question back in, and then another, and another, until we finally ended up in some steady-state with a question about finding nude photos of wrestlers. Ahem.

Via con Dios, Jeeves.

(*) For the nerds: yes, I knew what peephole optimization was, but the compiler seemed to be doing an awful lot of code movement, so I thought I'd read more about it. It actually turned out to be an issue with the build system, not the compiler -- the wrong source was getting compiled.

 

February 10, 2006

Net Neutrality

The Senate hearings on network neutrality were going on this week. I'm hoping like hell that Congress understands the ramifications of mucking with the network, but testimony like the following [pdf] (from Vint Cerf, now at Google, like everyone else) may not be particularly persuasive to most Senators:

Network neutrality need not prevent anyone – carriers or applications provider – from
developing software solutions to remedy end user concerns such as privacy, security, and
quality of service. The issue arises where the network operator decides to place the
functionality in the physical or logical layers of the network, rather than in the application
layer where they belong. Such a move is contrary to many of the fundamental architectural
principles of the Internet. In particular, attempting to solve applications issues at the physical
layer violates the layered, modular nature of the Net. With a few very narrowly-tailored
exceptions – such as defending against network-level denial of service attacks or router
attacks – altering or blocking packets within the network is inconsistent with the end-to-end
design principle. The end result is the insertion of a gatekeeper that – even arguably under
the best of intentions – disrupts the open, decentralized platform of the Internet.

It's amusing to picture the Senate getting down with the 7-layer model. I would love to flip on CSPAN-2 and catch a Senator saying, "Mr. Whitacre, I really don't understand why AT&T insists on making this a layer 3 issue, when this is clearly a layer 7 issue!"

February 06, 2006

Nicely done

It looks like GTalk now lets you save chats to your GMail account, something I brought up in a post a year ago. This was a smart, easy way to increase the value of GTalk.

I think Google's best bet for GTalk adoption, though, is to push the bot angle -- that is, turn GTalk into a platform as they have with many of their other offerings. AIM, MSN, and Y!IM all try to discourage end-user bots, which is truly foolish. It creates an opportunity for Google to introduce bots to the general IM public, who will think that Google invented them and will rush to switch.