UpdateThis column appeared in May, 1998. As often happens in the digital world, it wasn't long before almost everything had changed. The original column appears below, but please note that as of early 2000:
Search tipsYou'll get more from your searches if you pick two or three engines, and
familiarize yourself with how they operate. On some, you must type a plus sign
before each word: to search for gay hotels in Paris you'd type: |
From Yahoo to Webcrawler, from "gay Thailand" to "lesbian vacations": Which search engines are most useful for gay and lesbian travelers as they surf the Web?
Above: Sasha Alyson found enormous differences between popular search engines, when they were asked to search for gay travel topics.
The author: Sasha Alyson is the founder of Alyson Publications, the country's leading publisher of gay and lesbian books. He sold that company in 1995 to start Alyson Adventures, which specializes in active and adventure vacations for gay men and lesbians.
Table of Contents of Venturing Out columns |
Experienced web surfers are accustomed to this. The problem is compounded for gay and lesbian users, however. That's because much of what's on the web is porn-for-pay, and those porn sites work overtime to pump up their ratings with the various search engines. Once you type the word "gay" or "lesbian" to help define your search, you open the door to these hordes. Fortunately, some search engines prove smarter than others at detecting which sites are most relevant to your inquiry. After too many frustrating hours of watching search results download, only to find a list of spam, I undertook to test them. I took eight queries -- word combinations such as "gay hotel paris", "gay hike", "lesbian vacation" and "gay cruise ship" -- then called up the major search engines, and looked at the first ten results provided by each engine to each of my queries. If you think I must have been bug-eyed by the end of all this, you're right. I scored the results by asking, of each site listed, "How relevant is this likely to be for the typical user who types in the same query?" In response to the query "gay cruise ship", a listing for RSVP Travel received a top score. The site of a gay travel agency that sells cruises also got points, but not as many. Porn, spam, and irrelevant sites lost points. So did the home page of Mark, who is gay and writes of a memorable night on a cruise ship. Good for you, Mark, but that page doesn't belong in the top ten listings. And the winners are: InfoSeek (www.infoseek.com) and Search (www.search.com). came out on top -- a surprise to me, as I had barely heard of either one. They share a database, but InfoSeek consolidates all pages from the same site into a single entry, yielding a more concise list. Both did an excellent job at filtering out spam, and getting most of the best sites. Alta Vista (www.altavista.digital.com) was strong all-round, but especially with hotel listings. HotBot (www.hotbot.com) found more right-on-target sites than any other engine, but was weaker at filtering out the porn and spam. Lycos (www.lycos.com) was generally strong, but weak at finding hotel information. Yahoo (www.yahoo.com) actually hires human beings to look at each site before it gets listed. (Others use so-called "artificial intelligence" software.) Consequently, for most topics, Yahoo produced fewer junk sites than anyone else. Yet occasionally, your subject just doesn't fit into their scheme: I had no luck tracking down gay cruises, for example, on Yahoo. It also has a smaller database, and missed many relevant listings. Those humans go home at night, while computers keep on analyzing. Excite (www.excite.com) and Webcrawler (www.webcrawler.com) got rock-bottom scores. These turkeys share the same database, but apply different search algorithms, and so yield slightly different results. But "Garbage In, Garbage Out," as they say in the computer business, and both these search engines proved virtually useless at producing relevant listings. (It was Excite, by the way, that came up with the odd response to my search for the "gay hike" query. The porn site that got #1 listing includes the phrase, "If you are offended by the gay lifestyle, homosexuality, images of naked men, gay sex, or any type of sexual talk, take a hike. Leave now." None of the higher-ranked search engines made the error of believing that this site was about hiking.) Two major search engines were not rated: GoTo (www.goto.com) is undergoing a major overhaul. Commercial sites can now pay to get a top listing; GoTo claims this will help increase relevancy. They could well be right; it's worth trying them out in a few months. Northern Light (www.nlsearch.com) is heavily oriented towards news and text. It's not the place to look for gay vacation ideas; for current events, however, it's the place to start. A table summarizing these results (along with links to each engine) appears below. My conclusion: No one search engine will find everything you want. Get in the habit of using any two or three, from the top six. If you've been relying on Excite or Webcrawler, and could never understand why other people found the web so useful and you didn't -- well, now you understand. May, 1998 |
| Search engine | Score | |||
| InfoSeek | ||||
| Search | ||||
| Alta Vista | ||||
| HotBot | ||||
| Lycos | ||||
| Yahoo | ||||
| Excite | ||||
| Webcrawler |
Explanation of ratings:
I listed a number of word combinations that a gay or lesbian traveler might use: "+lesbian +vacation" or "+paris +hotel +gay", for example. For each combination, I made a list, in advance, of one or more sites that were highly relevant to the inquiry I had made, and that had been in existence at least six months. Then I typed this query (altering the format as appropriate) into each search engine, and looked at the results.
Found key sites rates how well a search engine found these sites.
Found relevant sites indicates how well it produced other sites that would be of likely interest to someone else making the same inquiry.
Filtered out spam shows how well it kept out irrelevant listings. Irrelevant listings may include deliberate spam (sites loaded with irrelevant words -- often porn sites -- in hopes of getting listed by search engines; accidental by-products (some search engines can figure out that an event in "Gay Head, Massachusetts" isn't of special interest to someone using the keyword "gay"; others cannot); and a few true mysteries (listings that had no apparent connection to my query, even when I downloaded them and examined for hidden keywords.) No search engine is perfect at this, but with its use of human editors, Yahoo! does the best job.