Sure signs of a search engine that hasn’t focused enough on the basics: when I tried Searchme today (after a long absence), and was surprised to find that loading the homepage stalled for dozens of seconds–showing nothing–while the bottom corner of my browser announced that it was loading Google Analytics scripts.

Sure enough, looking at the source for their home page, the first thing that happens in the body is the loading of those Google Analytics scripts.

Response time is incredibly important in search, and I am really surprised that they haven’t done the homework to pin it down.   If you want people to use your search engine a lot, you need to make sure that the homepage renders quickly, and focus is set to the input box really quickly.

Hunch from Flickr co-founder Caterina Fake is a “decision engine”.  After asking you some 20 questions to get to know you, you get to pick questions to ask it.  It then asks you a series of multiple choice questions to determine your preferences before giving you recommendations.  Examples of a top-level question (called topic on Hunch) are “Where should I go on vacation?”  and “What should I plant in my garden?”

The interesting thing is that all these topics are created by people — so in a sense it is the ultimate Web 2.0 search engine.   I found that it worked quite well in recommending designer suits for my wife and restaurants near my area.  Where the system breaks down is that it doesn’t take into account the fact that certain criteria are more important than others.  For instance, I played the topic “Should I get an iPhone or a Palm Pre?”  One of the questions was if I wanted to be able to multitask.  If this need completely outweighs all other considerations, then the answer is dead simple, and no other questions are needed.  But there is no facility for that.  So in its current form, it works best for “fuzzy” or subjective decision-making.

This puts it in the same market as Aardvark, which lets you ask real people subjective questions on IM, the difference being that with Hunch, the answers are pre-canned, and it is supposed to learn about your preferences as it goes.

I’m guessing they thought he was a spammer of some sort building up influence and thus a valuable Facebook account.  But if they would only dig more deeply, they would see that all his friendships are legit.  Bad algorithm.

It is great that the Shen Yun troupe is showcasing beautiful Chinese dance and culture to the world.  It is unfortunate that this is being done in conjunction with religious and political propaganda.  One of the dances depicted a Falun Gong practitioner being dragged off by black garbed agents with big red communist insignias on their backs (instantly raising the Pavlovian response in this audience of “Communist=Bad”).  He is brought back dead, but fairies and angels descend from heaven and take him into their midst.  I am not an apologist for authoritarian governments, but this sort of heavy handed cheese is really bad.

Then there was the male soprano who sang a song urging people to seek the truth with Falun Gong.  The lyrics were projected in both Chinese and English, and had obviously been dumbed down so as to have the widest appeal, but the dumbing down was disastrous–every line made me cringe.  That was all so horrible until I noticed that Gary had nodded off serenely beside me, while behind me, a fat, massive man was meditating to the song.  Now I really enjoyed that.  That was the highlight of the whole performance.

http://similar-images.googlelabs.com/

Works really well.  A few of its false positives are entertaining though: for instance, among those similar to the first hit for Chinese actress Zhang Ziyi was another actress Zhang Jing Chu…not a bad way to gain some fame. 

<———>

And how about this one. This guy is similar to Buddha…

<———> 

The fact that they both have big hair proves that the algorithm is doing visual recognition.