Search This Blog

Yesteryear

Saturday, April 5, 2008

April 5, 2008


           I’ve also begun my quest for stress-free work, if such a thing exists. One of the first places to check was the used book stores. That would be ideal for me. Same everywhere for all the places I checked had a lineup for the job. I’m reminded of the definitions of unemployment, like frictional and structural, but I believe this move would put me in the category of underemployed. There are very few such stores left any more. My favorite job over the past years were those weeks at the Thrift.
           A show of hands, who knows anything about SEO? Search Engine Optimization. I bought a ‘dummies’ manual which I predict will have no surprises. It’s worth reading and if you spotted most searches bring back consistently bad results, you are right. Almost all search sites use the same few systems (the top of the list is Google). All of them have trouble with polysemy (yet another word not in the Big Dictionary, so I hope I’ve used it correctly). These systems cannot find which of the nuances of “jack” or “bank” that you want, so they return them all.

           Also, none of the major search systems can do adequate searches on compound phrases like “post office” and “dog pound”. Often they are worse than no search at all. (Wankers searching Paris Hilton are not looking for foreign hotels.) Another problem area is punctuation, is that e-mail or email? Other than having a human create the indexes (search “engines” search their own indexes, not the Internet), some form of Artificial Intelligence is required before the situation can get any better. Were we not assured back in 1985 that this was “just around the corner”? May I quote the person who said instead of Artificial Intelligence, we got Real Stupidity.
           I wonder if AI might not have done better had not the Internet diverted so much attention away from other software innovation. One of the worst ideas ever is how search engines, if they can’t find the exact phrase, will try to find instances where the two words are physically close to each other in the text being searched. That’s a super-duh that clutters up the results page with nonsense. For some mystical reason, you cannot disable this feature, and lately I’ve noticed that using quotations marks is failing more often. Again, my solution is to use the square brackets to indicate literals, [like this] which I’ve been advocating for years to no avail.

           Later. During a very loud thunderstorm starting at 4:45 A.M. the morning after today, I used the time to read 104 pages of the SOE text. Very interesting. Not the book (which merely showed that I had already guessed 9/10ths of the correct techniques) but the insight into the things that don’t work for search engines. The very structure encourages constant warfare between users and abusers for top position. I still see the best short-term solution as allowing compound word searches and defaulting to an AND search (from the current OR search).
           What? Okay, I’ll explain that. Boolean logic. Right now, if you search on more than one word, such as “dog pound”, you find every instance of either word, that is, your search is interpreted as “dog OR pound” and returns millions of sites with one word or the other. If you searched only for articles with “dog AND pound”, you would at least stand a chance of finding something relevant. It would seem easy to allow searches on exact phrases. The search engine people appear horrified by the concept.

           I am aware that the advanced Google search has a text field for exact phrases but the only way to refine that search is by adding OR terms. Try it. It was good idea that shot itself in the foot. Very few web pages contain the exact phrase “Real Estate Florida”, but if you break up the phrase to the available fields of said search, you are back where you started with an OR. Google works, but it is still far from a good search system.
           The search mechanism itself is bad, so tweaking it to get higher on the search lists seems a little kookie. It does, mind you, explain why you can never find what you want with a search. Google, which dominates the search field, refuses to allow any type of categorization of web pages. Instead it uses a formula that counts the number of links to a site. Anybody who finished high school knows that popular rarely means smart. The more dumbbells that visit a web page, the more likely your search will land there.
           My favorite Google from 1998? A search on “rocket” brought back a song by Elton John.

++++++++++++++++++++++++++
Return Home
++++++++++++++++++++++++++