Well I'm always happy to see a new interesting web site, and especially something that seems to hold promise of something "different". So I welcomed the announcement of the new "Cuil" search engine. Understandably there are (and will continue to be) lots of startup glitches, like crashed/overwhelmed servers, but I have a much more serious problem.
One of Cuil's features is that they display a small image near many site listings in an attempt to make the listing look more interesting, and hopefully show something visual about the site contents. Selecting an appropriate image from each site is challenging, but that's their problem.
To test how this works, I did a search for my own name. I'm a professional photographer, so besides for my own web site, I'm listed in a lot of various directories, etc. Imagine my shock when I see a small portrait displayed beside MY WEB SITE listing that I DID NOT TAKE, and that DOES NOT APPEAR ANYWHERE ON MY SITE!
This is more than a mistake. Cuil is representing my business with an image that isn't mine! Two thoughts (at least) come to mind: 1) that if this is not a good portrait, or simply not my style, then Cuil is misrepresenting my business! and 2) Whoever does own that image, assuming they are also professional, may be very upset to see their copyrighted image displayed next to my site listing! (They could even charge ME with copyright infringement!)
The same can be said about some of the other directory listings that appeared in Cuil's search results. While the text portion was relevant to my business, some random photo from the site was displayed beside my information! This MUST BE STOPPED.
On a more minor note, Cuil's search algorithm seems very strange if not obsolete. One might expect that when searching for my name, my own site would appear first, as it does on almost every other search engine. On Cuil, it's about 9th (if you go top to bottom) or 3rd if you read across. It appears to be a simple context text matching algorithm... so if my name is mentioned twice on a directory/index site and only once on my own site main page, the directories appear first. No exactly best match in terms of user objectives in searching!
I can see lots of potential law suits here if Cuil does not very quickly address the image problem. I spend over 30 years in information technology, and can easily see that selecting an image is a next-to-impossible task. For example, I have several sub-sites (completely different topics, but hosted as sub-domains under my site). While this is perfectly legitimate, a search engine has no way to tell that the images under one sub-site are completely irrelevant to another!!
My suggestion? Either drop the image idea completely, or (better): offer to display any image names "cuil-image.jpg"... that way they can let the webmaster decide what image would be most relevant to their site. Cuil could still make a decision as to whether or not to display the image, but if they want to display one it would have to be one with that pre-defined name. (Maybe the rule could include multiple choices for random selections, like cuil-image1.jpg to cuil-image9.jpg?)
I hope the developers of cuil will move very quickly to address this serious situation, which I could term "information corruption" -- since it misrepresents information that is already correctly represented on the web!
One of Cuil's features is that they display a small image near many site listings in an attempt to make the listing look more interesting, and hopefully show something visual about the site contents. Selecting an appropriate image from each site is challenging, but that's their problem.
To test how this works, I did a search for my own name. I'm a professional photographer, so besides for my own web site, I'm listed in a lot of various directories, etc. Imagine my shock when I see a small portrait displayed beside MY WEB SITE listing that I DID NOT TAKE, and that DOES NOT APPEAR ANYWHERE ON MY SITE!
This is more than a mistake. Cuil is representing my business with an image that isn't mine! Two thoughts (at least) come to mind: 1) that if this is not a good portrait, or simply not my style, then Cuil is misrepresenting my business! and 2) Whoever does own that image, assuming they are also professional, may be very upset to see their copyrighted image displayed next to my site listing! (They could even charge ME with copyright infringement!)
The same can be said about some of the other directory listings that appeared in Cuil's search results. While the text portion was relevant to my business, some random photo from the site was displayed beside my information! This MUST BE STOPPED.
On a more minor note, Cuil's search algorithm seems very strange if not obsolete. One might expect that when searching for my name, my own site would appear first, as it does on almost every other search engine. On Cuil, it's about 9th (if you go top to bottom) or 3rd if you read across. It appears to be a simple context text matching algorithm... so if my name is mentioned twice on a directory/index site and only once on my own site main page, the directories appear first. No exactly best match in terms of user objectives in searching!
I can see lots of potential law suits here if Cuil does not very quickly address the image problem. I spend over 30 years in information technology, and can easily see that selecting an image is a next-to-impossible task. For example, I have several sub-sites (completely different topics, but hosted as sub-domains under my site). While this is perfectly legitimate, a search engine has no way to tell that the images under one sub-site are completely irrelevant to another!!
My suggestion? Either drop the image idea completely, or (better): offer to display any image names "cuil-image.jpg"... that way they can let the webmaster decide what image would be most relevant to their site. Cuil could still make a decision as to whether or not to display the image, but if they want to display one it would have to be one with that pre-defined name. (Maybe the rule could include multiple choices for random selections, like cuil-image1.jpg to cuil-image9.jpg?)
I hope the developers of cuil will move very quickly to address this serious situation, which I could term "information corruption" -- since it misrepresents information that is already correctly represented on the web!
Labels: Cuil, web search
posted by Hershel Belkin | 9:49 AM
