Slashdot Mirror
Google's Weakness, AltaVista's Strength
Some people love the results they get at Google, others are often disappointed. To a large extent, both the pluses and the minuses derive from Google's ranking system, which (as the folks at Google explain www.google.com/technology) depends largely on the number links to a particular page and the relevance of the content on those linking pages to the content on the target page, and the quality of the pages doing the linking.
Thanks to that complex and brilliant system, over time, the best pages often rise to the top of search lists. But that takes time -- a lot of time.
It works great for old, established sites to which many other old, established sites have linked. (It works great for my site :-) www.samizdat.com ). But new sites, regardless of the quality of their content, get short shrift. It takes 2-3 months for the new pages to get into the Google index. Then it takes time -- perhaps years -- for other "important" sites to discover the new site and link to it; and then months more for the new versions of those pages with those new links to get into the Google index.
So if I'm looking for content that is likely to have been on the Internet for a year or more, Google is great. But if I'm looking for fresh content, I'll go elsewhere.
For me, for years "elsewhere" meant AltaVista -- for two reasons. AltaVista used to add new pages to its index, for free, within two days of submission, while other search engines typically took weeks or even months. That meant they had the freshest content. In addition, AltaVista provided you with a set of very precise commands that couldn't be matched anywhere else.
Over the last year, as AltaVista has struggled to become profitable, they have destroyed their beautiful free submission process, trying to force Web sites to pay for submission. Free submissions (which typically come from the kinds of content-rich sites that I'm interested in) now seem to take three months or more -- no better than the other search engines and often worse.
Fortunately, the powerful commands remain -- for instance, the ability to exclude as well as include terms in your query. AltaVista lets you use minus signs and plus signs to indicate what you really don't want and what you do want. And for some specialized searches the exclusion is essential.
For instance, say you want to know what Web pages outside of your own site have links to your pages. At Google, I can do a search for link:samizdat.com or get the same results by going to their "Advanced" search and using their "page specific search" to find pages that link to a particular page. But my results are then littered with pages from my own site -- information I don't need and don't want. At AltaVista, I can search for +link:samizdat.com -host:samizdat.com and get exactly what I want -- finding out who thinks enough of my pages to have linked to me without my having contacted them: a valuable list of well-wishers and potential partners.
Similarly, Google lets me restrict a search to a particular Web site. For instance, if I include in my query the term site:samizdat.com or in Advanced search under Domains I choose to restrict the search to that domain, Yes, I get results only from that site. But to use that command, I need to have additional query terms: site:samizdat.com alone generates no results.
At AltaVista, however, I can search for host:samizdat.com and get a complete list of all the pages at my site that are in the AltaVista index. Or I can search for url:samizdat.com/isyn and get a list of all the pages in that directory at my site are in the AltaVista index. Or I can search for url:samizdat.com/consult.html to see if that particular page is in the index.
In other words, AltaVista provides a higher level of precision and the ability to get information that is particularly valuable to people in charge of Web sites and Web-based marketing projects. And if they'd just fix their free submission process and provide the service they used to, they'd kick Google's ass for searches for current information.
P.S. -- The folks at Google are very proud that their system defies human tampering. In fact, what they've done is encouraged the development of bizarre business models structured to take advantage of their link-based ranking system. For instance, Webseed Publishing now has over 1000 sites, all with different domain names. These content-rich sites are each run by different dedicated individuals. (I'm one of them :-) In many cases, the content deserves high rankings for its quality. You might wonder why the umbrella business for all these sites bothers to maintain over a thousand different domain names, when it would be far simpler and cheaper to have them as directories under a single domain. But because the domains are different, the many thousands of links these sites have to one another all count toward the automated calculation of their popularity and quality at Google, giving them all a boost in the rankings and hence bringing Webseed more traffic and hence more revenue.
P.P.S. -- AltaVista appears to be making a comeback. Six years ago, when I was in the Internet Business Group at Digital and Digital owned AltaVista, about a third of the traffic to my Web site came by way of AltaVista. Whenever AltaVista had a glitch, I saw it immediately in my traffic stats. In fact, I sometimes was able to alert the engineers at AltaVista about problems before they had noticed them themselves. Over the years, due to increased competition from other search engines and also due to the business folks at AltaVista making bad decisions and jettisoning great capabilities/services (like 2-day free submissions, their affiliate program, LiveTopics, and newsgroup search), the number of people finding my pages by way of AltaVista plummeted. By January 2002, only 1% of my traffic was coming by way of AltaVista, despite the fact that as a long-standing fan and also as co-author of the book The AltaVista Search Revolution, I had lots of information about AltaVista at my site. I was actually getting twice as much traffic from the International Atomic Energy Agency (part of the UN), when I had no information at all related to atomic energy. But in recent weeks the traffic from AltaVista has climbed sharply. It now amounts to 6% of my total. I wish I knew why that was happening. In any case, I hope that trend continues.
I'm amazed, an entire article astroturfing AltaVista. Sadly, the author is a bit short-sighted, and doesn't realize how quickly stuff appears in Google's cache (often within weeks, less than a month), or that even if something accidentally ranked lower because of the number of links a given page receives, it still ends up in the first page or two anyway. *Sigh*
i wish more sites would develop tool bars similar to google... it is extremely convienient.
on all my windows boxes it is one of the first things i install.
google is probably the best search tool right now, and they make using it a breeze. altavista used to be the best search tool, but they made it harder and harder to use, and then search tool lost its top spot. totally different situation. if google looses its top spot in the search tool field, i'll still use it for its ease of use.
MARIJUANA, SHROOMS, X: ONLINE?! - E
Boolean mumbo-jumbo? That's the best PART of AltaVista. Google limits querys to 10 words? That stinks! Google is great for simple querys about common subjects. AltaVista's boolean query is great for finding that site whose link you can't remember but you remember some of the words that were close together. AltaVista's boolean query is great for finding information on little-known subjects that you can pretty well guess what keywords will be near each other. I used to use AltaVista's boolean query exclusively. Now, I find it's best to try both AltaVista and Google. Each find content the other won't.
Of course you can find things with search engines now. Google's "trick" of counting links helps a little bit for a particular class of query, which is when you know the name of an organization and you want to find its site...it works well because more people will link to the site as opposed to other sites that discuss it. But as I have written elsewhere, if AltaVista is 99% lame, then maybe Google is only 97% lame...which is three times better, but still terrible if you take a step back.
Now Google is doing a lot of good things outside from its basic search engine, which should be applauded. The caches, saving old Usenet posts, the image and catalog searches, etc. are all good things -- but they don't affect its basic ability to search well.
Further karma ho' expounding can be found right here.
- adam
... because it is so good.
...
I'm a librarian. It is the most difficult time in history to do library research. There are hundreds of overlapping commercial databases out there, each with their own coverage, interface, and search engines.
Students used to locating information with Google are appalled at the steps it takes to locate a scholarly journal. You need to browse a list of subject databases, search them, then locate a printed copy of the journal via our catalog (a growing but still small percent of journals are available online).
Someday searching the various literary databases may be as easy as Google, but in the meantime there are drastic capitalist impediments to making it easy to do library research.
... so ask a Librarian if you ever need help
I would like to see a program and specification that dictates a formal data format for information in a mathematical schema. This could be the foundation for a universal translator and certainly a decent means of doing a search engine.
The idea is pretty simplistic, although the implementation is complex.
Any communication takes place by translating an idea into a sensory input form.
Examples: Sight (written language, video, sign-language), Touch (brail, texture), Sound (conversation, music), Taste (Like water for chocolate?), Smell (pheromones?).
Obviously, not all of these mediums are easy to work with, but we can certainly start with written language.
All languages use the same basic principle: convey relevant information about a central subject. How they go about doing it is different even between versions of the same language (British English vs. American English).
If we described an objective hierarchy of physical objects described by pure mathematics and implanted them into a central, world-wide database then open-source parsers for each language could handle the task of translating any written text, in any supported language, into this common language. If correctly implemented a search engine could enter into a short dialogue with a person performing a search and then return information very specifically relevant to what the user was searching for.
Example dialogue:
[user]I want information on Mary Jane Carpenter.
[google]There is a very famous person by that name. Her official website is [here]. [Here] is a list of fansites and [here] are some other sites which discuss her. That name is mentioned in [these] sites, but it is unclear if they are talking about the same person. [Here] is a list of other people with that name.
[user]The person I am looking for isn't famous.
[google]Then you are probably looking for one of [these] people.
[user] Are any of those people from St. Lewis?
[google] [Here] is a sight dedicated to a Mary Jane Carpenter from St. Lewis.
This may sound like an impossible streatch but it really isn't. The famous Mary Jane Carpenter has a unique id on her object and many thousands of attributes which uniquely identify it from any other Mary Jane Carpenters. Ambiguity is dictated by the same rules that govern conversation: context.
If I have a page that contains no content other than Mary Jane Carpenter sucks! then a simple fuzzy logic routine should be able to infer that the Mary Jane Carpenter I am talking about is probably the famous one. Other clues could be gained from other parts of my site or other documents which have me as a source.
I realize that I am talking about a HUGE database, but it sure would be handy...
My $0.02 will always be worth more than your â0.02, so
And that is that I sometimes NEED to use the near keyword in altavista to get a complex search to work correctly. If google added the near keyword I would get rid of my quicklink bar entry for altavista's advanced search.
p.s.
the advanced search page is all text, not even a banner ad so it's almost faster than google to load.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
According to http://www.leekillough.com/robots.html - iaea.org is commonly used as a fake referrer by spam harvesters.
My amazing wife - Artist, Author, Philosopher - Laurie M
As it was over-explained, Google ranks pages according to how many links elsewhere points to that page.
:D
Remember this post from Slashdot ? It is about Macromedia wanting Flash to be used to design the entirety of a site.
So, I don't suppose Google can fetch the URLS inside a Flash file (correct me If I'm wrong), so, if Macromedia's dream become true, how would Google cope with it?
BTW, how any search engine would deal with such a catastrophe?
Cheers.
I type in google.com and it loads damn near instantly.
The Google folks were at a local user group meeting a few months ago. They told us that they have byte counters -- the human kind -- monitoring how many bytes each page served takes. Their mission is to keep the count down.
They got very noisy applause for that statement.
Andrea
While you are right about people stressing the wrong differences, your claim about AV and Google's respective syntaxes being interchangable is false.
In AV's simple syntax, unadorned terms are not required to be in the results, they are only preferred. If you want to require a term, you must prepend a plus sign. Google does not allow you to distinguish between required and preferred terms. This may arguably be simpler, but it takes away control from you, the user.
AV also offers a separate advanced syntax which provides support for nested booleans, and positional operators like "near", "within", "before", and "after". Google, while it allows a single level of simple booleans, does not provide any means by which to nest them. It completely lacks positional operators other than phrase matching. Again this takes control away from you, the user.
It never fails to mystify me why Slashdot readers, a crowd biased strongly towards programmers, engineers, and Unix users, namely people who love to have lots of control over things, would favor a dumbed down search service! (I agree that there are other problems with AltaVista, such as the annoying popunder ads. However, these have nothing to do with the quality of its search syntax.)
But my grandest creation, as history will tell,
Was Firefrorefiddle, the Fiend of the Fell.