Find it on the Web

One fun part of the “Web guy” job here at Milk Pail (dot-com, cha-cha-cha) is getting to see what works — and what fails risibly — for average Websurfers doing a little casual research. [Somehow, I suspect our big melons weren’t quite the type one hapless Internaut went searching for. Hey, that’s showbiz.] So, in the spirit of the Web’s early days, I thought I’d share some of what works.

Rule #1: Spell it right! Today’s search engines are somewhat tolerant of misspellings, but in trade for being allowed to be a little off-target you get results that are way off-target — near-match spellings with entirely unrelated meaning. Probing with correctly spelled terms pushes your best results to the top of the list. (By the way, MSN Search seems to apply fuzzy matching to the topic: they sent us someone who looked for “Chinese astrology”. Well, we do talk about sheep and goats.)

Corollary: When you can’t decide between two possible spellings [or two usage variants], try Web searches on each and see who has the votes. Given the endemic decline of literacy, you won’t always get a technically correct answer, but you’ll at least have a widely accepted one. :/

Odd is good. There is absolutely no point in including a common word like [Internet] in a search query, even if intimately related with what you’re looking for. Search instead for the most unusual term that’s expected to appear on the Web pages you want to find.

e.g. [turunmaa] beats [cheese] if you’re looking specifically for Finnish cheese.

Be short and sweet. Using too many search words may end up knocking out pages you wanted to see:
[is there a target store in the sawgrass mills mall] may possibly work, but [target sawgrass] almost certainly will work better.

And if it weren’t for random personal web pages, both searches would fail for a different reason:  the official Target site makes up store-location pages only on request — they have no page (just now) with both the words “Target” and “Sawgrass” for Google et al. to find and index.

Use a specialty database. If you’re looking for that little parable about cheese that got mov*d, why not start at some “Amazing” online store that has a database of  just book titles? Yes, you then miss the serendipitous side trip to MilkPail.com, but we’re on a mission ;)

Interestingly, this approach fails just where you’d think it would shine: finding people in the phone book. Some of the reasons: unlisted numbers (about half of urban Californians are!), obsolete data, uncertainty about the town or state, and a profusion of same-named persons you’re not trying to find. You might be better off searching with Google for the (quoted) name together with a word related to the person’s interests or occupation: ["Steve Rasmussen" cheese] More often than you’d expect, you’ll hit a “bingo” with a résumé, personal Web page, or even a forgotten guestbook/bboard posting. It’s all fair game here. (Hint: Don’t overlook deja.com for possible posts by/about your quarry on Usenet.)

Superior content earned Milk Pail a high ranking.

Listen to the experts. If your topic of interest is fairly broad and well-known, try one of the humanly compiled “tree”-style indexes such as dmoz ODP or LookSmart. You’ll likely find links to several good general background essays on the Web. For concise surveys of a topic, you might also try Britannica or Wikipedia.

Use a high-powered tool when you need one. Google does a lot of things right — huge database, quick response, good at ranking for relevance, and back-up copies [of all “8,058,044,651 web pages” !?] so you can even read stuff that’s been pulled down from its original home. And the kicker is, with the in-context excerpts that come back (isolating the part of the page where your query terms appeared) you often get the info you needed without even clicking through to the found page.

Devil in the details. You can improve the precision of your results by quoting word groups that must appear together in a ["particular sequence"] and by capitalizing anything that would be capitalized on the ideal page you’re looking for. For example, search on Brad and [at least on some engines] you won’t get stuck [ow — bad pun] with pages about fasteners; search on ["Brad Pitt"] and you won’t get pages that just coincidentally mention Brad Smith and Olive Pitt. (Another tip, by the way: don’t even bother looking at pages larger than about 100k bytes — they generally have so many words, they can match just about any query without being remotely related to the subject you had in mind — and worse, they likely exist just to push malware.) And if your search involves someone as famous as Brad Pitt of the movies, you’d be wise to narrow it with qualifiers like, say, ["Fight Club"] so you’re not slogging through every idiot thing written about the man.

Don’t forget to check your favorite search tool for help with its “advanced” options. You may be able to restrict your search by site domain, file type, or other useful parameters.

Close isn’t just for hand grenades. Sometimes a page that seems barely on-topic, points to a really helpful page. Or it may suggest a particularly effective word to use in your search. (An “effective” word is one very likely to appear on the kinds of pages you’re after, and very unlikely to appear on other pages.) If your ’Net connection is decently fast, give a quick look to these second-tier pages. Use common sense though; if the Google excerpt of a page is just a heap of words, not even coherent sentences, someone may be trying to game the system — and pull you into a tenacious vortex of ad pages and malware.

“But this page isn’t even close.” If your results page looks hopelessly off-topic, your search terms probably really are in there somewhere — use your browser’s Find In Page function (Ctrl-F/Clover-F) to see just where — or else a really close synonym fooled the search engine. Google has some excellent tips on dealing with this and other “advanced” search issues. (Hint: The plus sign is your friend.)

Ask the answer, not the question. ["snowball rolling"] may help you more than [Temptations lyrics] because you won’t end up with pages (this one, for example) that talk about Tempts lyrics but don’t cite any.

The “ask the answer” strategy works even better when you parlay it. Say you’re trying to find all the bowling centers in Broward County* (Fla.) and even the Yellow Pages only shows one (!). Try taking the name it does show and plugging that into Google to find some more. Repeat as necessary with every new name you find :)

Use a meta-engine. Run your search on Dogpile or SurfWax, which will pass it on to a half dozen leading engines.

Think outside the (visible) Web. UC Berkeley has an excellent plan for expanding your online research horizons.

If you still come up empty-handed, maybe it’s really not out there — or if it is, nobody knows where. But console yourself that you gave it a good try, and in the world’s largest library at that. (If the elusive subject is one you really care about, consider putting up a reference page of your own once you’ve learned more. That’s what makes the Web work.)