Forum Moderators: open
One idea might be to start with your own directory...
- It should be free so as to not exclude sites without funding.
- The structure could be similar to Dmoz of today, at least as a starting point
- The site would be built by hand, by employees of Google, subject to quality control standards.
- Sites would be rated by Google on a 1 to 10 scale, similar to PR of today based on the overall quality of the website.
- Search terms would be mapped to various directory categories, with the strongest relationships given higher weights. For example, a simple search term of "car" might be mapped to automobile makers and online car dealers among others.
- Drill down would be presented to the searcher to clarify the search, for example Buying a Car, versus Researching a Car. This concept is similar to Gigablast's presentation of results.
I realize the directory is a problem to maintain, but I'm not sure how else you could insure high quality of results short of manually evaluation.
Thoughts?
Example? User types in word(s) and results 1-10 show up in some order for that word that no one fully understands and always leaves numerous websites and searchers at a disadvantage due to a mysterious algo that changes every month, er... three months, uh, six months? Of course, every SE works this way, not just Google, but I feel a lot of frustration for Google, site owners, and searchers could be solved by reconceptualizing they way the results for those words are arrived at and displayed on the screen. Everything on the screen is up & down. Your site is in the top ten or it is never found... who said the page had to be displayed this way? Just because that is how all the others do it?
It is time for a new paradigm in SERPs. Nash's economic theory of supply and demand comes to mind; everyone invariably rushes for the hot blond (position #1) insuring everyone goes home empty-handed. In other words EVERYBODY wants to be number 1-5, so some employ less than fair tactics to achieve that ranking (which punishes those who follow the "RULES" that Google doesn't want to quantify) and once the brainiac of the month at Google get's wise to the method being employed to manipulate results 1-5, Google changes the rules again and screws everything up for everyone and then we have the Florida update all over again.
So I ask, "Why does it have to be keyword to single-column SERP?" Why can't there results that further quantifies the keyword being sought, by industry, by location, by price? Are you looking for a company selling a product called "widgets" or simply information on what widgets are?
Here's my idea. Have the SERP page 4 columned. First column commercial, second column would be informational only, third would be geographically pertinent, and fouth revenue building (adwords, paid lisitings, banner, etc). This way there would be a total of 4 number "1-5"s on the screen and would allow users to understand what the insuing site will be before they click on it.
Whadya think?
Directory = an index of sites
Search engine = an index of pages
A directory might be handy if you're looking for sites that are devoted entirely to Elbonia. But if you're looking for articles about Elbonia, a directory won't point you to pages in nationalgeographic.com, newyorktimes.com, guardian.co.uk, etc. that may be of equal or greater value and relevance.
As for Analonda's suggestion that SERPs have multiple columns for informational, commercial, and other types of search results, I think that's an interesting idea but is impractical for two reasons:
1) It doesn't work very well in a narrow browser window (as anyone who's ever used the "Google Dance" tool can attest); and..
2) If the user is looking only for informational or commercial results, why clutter his field of vision with three other columns of irrelevant results?
A similar but less cumbersome approach would be to have radio buttons on the search page that would allow the user to weight the results according to his search preferences, e.g.:
"I want information on" and "I'm shopping for."
Exactly! I've long thought about building a directory that did precisely that - let people indicate whether they simply want information or if they want to buy something.
The all-in-one-basket approach doesn't work with today's diverse sites. We have blogs, ecomm, info, news, etc. sites. So let people
a. specify what type of content they're searching for
b. know what results they're going to get from a search
Makes for better (more relevant) results in the serps and more educated users as well.
Heck - why not institute a meta tag for "site type" based on those categories. I know one exists, but the SE's don't do anything with them, and most people don't include them...yet.
And speaking of "new paradigms." Why display 2 million results for a search query? Who's gonna look at them all? Index and rate them, but only display the top 50 results or so in the primary index. If users want to search the "supplemental index" they can choose to do that manually from another screen...and people can still check their rankings within the supp index.
Heck - why not institute a meta tag for "site type" based on those categories.
Because they'd be abused.
A directory can never be a "perfect search engine" simply because it isn't a search engine.
Directory = an index of sitesSearch engine = an index of pages
I mentioned that the search engine would START with a directory of its own. The directory itself would be used in a fashion similar to pagerank or the concept of Hilltop where authority would be determined by position in a directory and a quality rating of the the site.
In the situation of google, linking becomes meaningles. A combination of directory category, a qualitative assessment of the site and on page factors would be used to determine the SERP.
Once the results are returned, the user would have the ability to "clarify" what they are looking for in a way similar to Gigablast offers the user a drill down.
Spammers would just build an info site that should be in the "shopping section"
Possibly, but I think Google's PhD's are smart enough to detect ersatz "information sites." Of course, if a site that's trying to sell things also provides useful information, there's no reason why its information pages shouldn't be in information-oriented search results. Let's say, for example, that REI has a page on "how to buy a canoe" or Land's End has a page on "how to shop for cashmere sweaters." Such pages would clearly be relevant to a search on types of canoes or cashmere sweaters.