The Search Engine Update, Jan. 4, 2000, Number 68

January 4, 2000 - Number 68

By Danny Sullivan
Editor, Search Engine Watch

About The Update

The Search Engine Update is a twice-monthly update of search engine news. It is available only to those people who have subscribed to Search Engine Watch,

Please note that long URLs may break into two lines in some mail readers. Cut and paste, should this occur.

In This Issue

+ HotLinks: Harnessing Bookmarks For Search
+ RealNames Temporarily Suspends Registration Of Generics
+ Longer Domain Names Arrive
+ Search Engine Resources
+ Search Engine Articles
+ Subscribing/Unsubscribing Info

General Notes

Hello Everyone--

Happy New Year, New Century and New Millennium to you all. For the purists who say 2001 marks the new millennium, don't worry -- I'll send you well-wishes next year.

I've been getting a number of emails about when the next Search Engine Strategies conference will be, and I'm happy to say we should have a final date pinned down within a week or so. I'll post details on the Search Engine Watch home page, as soon as they are available -- or check the seminar site, URL below. It will probably be on March 9, in New York City. The format will be very similar to the one that was held in San Francisco. Seminars

With the holidays behind, I'm diving right into making site updates. You'll find the Search Assistance Features page now reflects some new capabilities at MSN Search, such as page clustering and stemming options. This is due to MSN's return to using Inktomi, as mentioned in the last newsletter, and you should also review other expanded options on MSN Search's advanced search page. The at-a-glance Search Features Chart for Searchers has also been updated for MSN Search and AOL Search. The Using RealNames Links page now reflects the implementation of RealNames at Google. Search engine ratings pages from Media Metrix and NetRatings have also been updated -- and they'll get another update later this month. Links to all these pages can be found off the What's New page.

What's New


HotLinks: Harnessing Bookmarks For Search

When I first heard about HotLinks, I was dubious. After all, this is a search engine based on using people's bookmarks -- and knowing the clutter and junk within my bookmark list, my assumption was that HotLinks would only magnify the mess.

Now having played with the system, I'm more optimistic about it. HotLinks is not going to replace your favorite search engine, especially when you need to do a comprehensive scan of the web. However, HotLinks is likely to develop into a useful tool for finding golden nuggets that larger search engines may miss.

Bookmarks form the heart of HotLinks. Over 100,000 people to date have opened accounts and imported their saved links into the HotLinks system. It's an easy process, and one that allows you to access your favorites from any computer

All that data is also used to power the new HotLinks Guide, which launched in September. The guide organizes many of the bookmarked links into categories similar to those used by directories such as Yahoo.

HotLinks has far to go before rivaling the major directories in depth, however. For instance, its Travel category is divided into three subcategories. In contrast, Yahoo has nearly 300 travel-related categories.

While HotLinks may not be comprehensive, the smaller scale might make it feel more manageable to some people. For instance, Yahoo has multiple categories stuffed with sites devoted to the Palm -- heaven for any Palm enthusiast. In contrast, HotLinks has a single Palm category that lists just three sites. But those sites are extremely good, and for someone who's looking to get started fast with the Palm, HotLinks probably gets them to the right places faster.

Unfortunately, I'd say this is the exception rather than the rule, for the moment. The service is still so new that oddities stand out. For instance, the second most popular site in the Vacation category is for a Wisconsin vacation homes site. I'm sure Wisconsin is a lovely place to vacation, but I find it hard to believe that this particular site would be so highly rated.

Hopefully, things will improve as the service matures. One potential strength is they way HotLinks ranks sites within categories. Sites are listed in order of "Guide Score," which is the number of bookmarks in the system that point to that particular URL. The more bookmarks there are to a site, the it ranks in a particular category.

In a way, this makes HotLinks similar to Google and other search engines that rely on link data to bring popular sites to the top of results lists. The difference is that instead of looking at links on the web pages, HotLinks makes use of links in bookmark lists.

It also means you might find some good sites that the other services might miss. That's because bookmarking is easy, in comparison to web page publishing. Using links from bookmarks might yield a different set of results than from using links on web pages, and it seems especially likely to bring up some sites that might be otherwise overlooked.

How's the organization done? HotLinks CEO Jonathan Abrams says that software is used that tries to find patterns to automatically assign sites to categories, where possible. Of course, the software isn't perfect -- so HotLinks also has editors involved in the process.

HotLinks isn't the only online bookmark service available, but it is notable for pushing strongly into the search space and for its backing by CMGI, which owns AltaVista and has a stake in Lycos. Thus, it wouldn't be surprising to see some important partnerships with HotLinks emerge in the near future.

How about being listed? The best way to ensure your site is represented is to bookmark it yourself, then open a HotLinks account and import your information. Potentially, if you were to open several different accounts, you could increase your popularity within the system. However, Abrams said there is monitoring of any suspicious activity.

A better strategy might be to encourage people to bookmark your site and perhaps even direct them to HotLinks, which offers an affiliate program. Some resources for this are below.

Overall, it's not worth a lot of time worrying about how your site appears in HotLinks. But by doing some basic preparation work now, you should be set in case it or other bookmark-based services rise in popularity.

HotLinks Guide

HotLinks Affiliate Program


Another online bookmark service, though it doesn't offer an associated web-wide search feature, as does HotLinks.

Backflip brings bookmarks to next level
San Jose Mercury News, Dec. 11, 1999

Good review of the BackFlip service.

Bookmark Us! Script

This JavaScript applet lets Internet Explorer users bookmark your site by clicking on a link, while Netscape users are told the browser command to use for bookmarking.

The Dastardly "favicon.ico not found" Error
Web Developer's Virtual Library, Aug. 3, 1999

If you bookmark Search Engine Watch using Internet Explorer, you'll notice the site has its own unique icon that appears next to its name. You can do this too, by reading the article above.


RealNames Temporarily Suspends Registration Of Generics

Last month, I reported how RealNames had been allowing some companies to register generic terms as RealNames Internet Keywords, a significant change in its previous policy against generics. Now RealNames has temporarily suspended the registration of generic terms altogether.

"There were multimillion dollar deals that we've had to suspend that we're not going to sign now," said RealNames CEO Keith Teare. "It was a big decision to make."

The company is reacting to a recommendation from its Policy Advisory Board that a formal approval process for generic terms is needed to ensure that the overall integrity of the RealNames system is being maintained.

The policy board became involved about three months ago, after a RealNames customer and major web site operator tried to register a generic term as a RealNames keyword. The company was told that RealNames policy didn't allow them to register the word because it was generic, but then a competitor was later granted the term.

(Note: I wasn't able to get the an OK on releasing the companies involved before this went out. If this comes through later, I'll update this article online to name them)

That the word was later allowed underscores the "confusion" over generics that RealNames has dealt with for much of last year, Teare said. The company began assigning more and more generic terms, in particular to companies that also owned the generic .com equivalent.

"We ended up almost accidentally allocating generics only to the .com owners," Teare said.

The policy board disagreed with the accidental policy that had emerged. "Generic .com companies and their non-generic .com branded competitors should be allowed the opportunity to vie for the right to have the generic category," its recommendation of Dec. 15 declared.

The board further outlined some main criteria that it felt should be considered by RealNames in determining if, and to whom, generic terms would be assigned:

+ Current user expectation.
+ Ability of the requesting company to maintain or build user expectation, as evidenced through the degree of marketing commitment the companies would be willing to undertake.
+ The financial proposal to RealNames for resolution of the generic term.

Of these three criteria, I'd only support the first -- current user expectation. For instance, I'd agree that most users currently would expect that the RealNames keyword "Amazon" should navigate them to the Amazon web site, not present them with a list of sites that might be related to that generic term. It makes prefect sense to evaluate other generics like this and assign them as appropriate, which is something RealNames has always done.

I find the second criteria alarming. It basically says that any company which can convince RealNames that it will spend a lot of money in an attempt to make itself synonymous with a generic term should be considered for that term. So potentially, if Amazon were to pledge to spend millions marketing itself as findable using the RealNames keyword of "books," RealNames might consider granting it that term.

This is particularly disturbing because the company would immediately benefit from the term, even it if hadn't yet earned the user expectation. Overall, the assignment of generic terms is tricky enough without RealNames essentially encouraging companies that haven't earned a generic identity to seek them.

That leads to the third criteria -- the fact that the financial deal proposed to RealNames should be taken into account. How much RealNames will make off of selling a generic keyword has nothing to do with whether that term is appropriate to be issued, and this shouldn't be part of any review policy.

The exact criteria that will finally be used remains for RealNames to develop. Teare said that he expects the company will have a generic term approval policy in place and be ready to consider generics again by this April. He also predicts the policy will mean RealNames will be less liberal about generics than it has been recently.

"Our feeling is that there will not be lots of generic keywords assigned, because of this policy," Teare said. Furthermore, "If a customer really wants it, we'll open it up not just to them, but also to others in their category." In other words, deals will no longer be cut quietly on the side. If someone like Amazon wanted to register "books," other companies such as Barnes & Noble would also be invited to contend for it.

Teare also stressed that money would not be everything, when it comes to granting generics. "It doesn't relate at all to how much is paid or even who is the highest bidder, but to the ability to communicate their Internet keyword," he said. That means that potential generic term holders would have to make a strong case that promote their sites using the RealNames keyword.

As for the deals already done, they'll remain, for the time being. "We have to honor existing contracts, but at the conclusion of those contracts, the keywords in question would go into the [approval” process," Teare said.

A final comment on this issue. The policy board's letter to RealNames makes a point to delineating between "search" vs. "navigation" events. Search is where you are presented with a list of possible choices, because it is impossible for the service to know exactly what you want. Navigation is where a person is taken to one particular place, and it's what RealNames is all about.

"Navigation is about going to one place, not to another list of choices. We support the company's position that to present a list when a generic term is entered would be contradictory to the company's mission and service value proposition," the board wrote.

What RealNames needs to keep in mind is that its main distribution method is through search engines. In those places, it does a great and needed job in helping them with navigational issues -- the RealNames link of "Amazon" taking you straight to, for instance. But when RealNames forces generic terms into a navigational role, it weakens the value of its system at search engines by reducing choice and confusing consumers indeed wanted a selection of sites.

This is something the policy board itself recognizes in its recommendation letter. "The transition in consumer behavior from search to navigation is not an easy one. It requires that the company make difficult choices that might have an interim effect of creating a certain degree of consumer confusion," it wrote.

Unfortunately, board seems to assume that consumers will ultimately learn to navigate, rather than search. That's not going to happen. People will always know brands and want to reach those brands by navigating to them. But likewise, people will always be in situation where they need a list of companies providing particular goods and services, which they will obtain by searching.


RealNames Slipping Generics Into Specifics
The Search Engine Report, Dec. 6, 1999

Discusses the mixed signals RealNames has been sending about generics, which are supposed to be corrected when the new generic terms policy is in place.

Using RealNames Links

Describes how to find RealNames links at major search engines, including Google, which added them in December.


Longer Domain Names Arrive

Several domain name registrars are now selling names of up to 63 characters long (67 if you include the .com, .net or .org portions), which has sparked a wave of speculation that these longer names pose new opportunities with search engines. I've been flooded with questions on this subject, so here's a look at the issues involved.

In the past, .com, .net and .org domain names were generally limited to 22 characters (26 including the extensions). Now that the limit has been raised, names that were never possible to register are available. From a marketing standpoint, the new names offer some possibly significant value. From a search engine standpoint, the value is negligible.

Let's start with marketing. The advantage of having a domain name based on a generic term is obvious -- it's easy to remember for those thinking of finding a site about that product or service. Sites like,, are just some examples of businesses that have been capitalizing on generic terms successfully, from a marketing perspective.

The longer domain names may offer a similar marketing advantage, to some businesses. For instance, "" is now possible. Potentially, someone could register that name, then hope that anyone looking for "Yosemite camping reservations" would run the words together and slap on a .com. Additionally, this company could run marketing campaigns and hope that the name would stick in people's memories.

Now for search engines. Does having mean a guaranteed rise to the top of search engine listings? Absolutely not. Could it help? Perhaps a little -- and only a little, in my opinion. I have looked at numerous cases where people are sure that having the terms in their domain names is what has improved their rankings. In all of these cases, I've found that there are plenty of other factors that are also coming into play. Nothing has convinced me that the URL text is a secret weapon, and you see plenty of examples of sites that continue to rank well without having keywords in their domain names.

I'm not trying to rain on anyone's parade but rather put expectations in perspective. It probably won't hurt to register some long domains that contain your keywords in them; it even might help a little, but don't expect miracles from it.

For further information, I put the question out to some of the crawler-based services, and here are the responses I got just before the holidays:

"Keywords in the domain name do not help much in ranking. We look at half a dozen factors in ranking. The words on the page, their frequency and position on the page, are still among the most important factors," said Don Dodge, AltaVista's Director of Engineering.

Inktomi said that it tries to list a site first in its results if the domain name matches the query and if the term is unique. For instance, look for Quokka, and it's likely will be first in Inktomi's results, as it is a unique word. But search for "sports," which is more generic, and you'll find that being "" is not particularly helpful.

"When the search term or the domain name is pretty generic, then the URL name becomes less important and is only a minor factor that influences the ranking of results," said Andrew Littlefield, Inktomi's Director of Search.

As for Go, "At this time we do not appreciably upweight matches on URL paths --- so having a very explicit domain will not help a site much," said Jan Pedersen, Go's Director of Search.

Still thinking of registering a new, long domain name? Then here's some practical advice, Q&A style:

Question: Do hyphens in the name help, such as vs.

Answer: Potentially, using the hyphens would be better, because they should break the domain name up into discreet words. Without them, you just have one long word. However, I would never register any hyphenated domain name without also having the single word one. It leaves you too vulnerable from a marketing perspective. You don't want a competitor obtaining the single word version and capitalizing on any brand identity you build with the hyphenated version.

Question: How much repetition of a term in a domain name is too much? For instance, would "" be rejected or penalized for spamming?

Answer: That's an actual question I've received, and the answer is unknown. To be honest, I don't even think the major search engines have considered this as an issue, since the whole concept is relatively new. In general, URL text isn't that important anyway, so I don't think you have to be concerned that there are any particular spam filters you'll fall into by having names like that.

Question: Can search engines crawl sites with long domain names?

Answer: There's no indication they'll have any problems. As long you can reach a site with your browser, a search engine should be able to do the same. Both AltaVista and Inktomi explicitly said they'll have no problems.

Some last issues to keep in mind:

While URL text isn't that big a deal, link popularity is. All major crawler-based search engines are taking it into account in a significant way. For those who own multiple domain names, this means you should chose one of them as the main domain that you'll be promoting online. Use this domain, and only this domain, for all your online publicity and as your "public" address. That will help build the popularity of that domain.

Any other domains you've registered should be kept "private." Let them resolve back to your main web site, so that you can capture any accidental traffic, such as people guessing at the domain name. But if you also promote these other domain names in a big way, you'll end up diluting the reputation that your primary domain can build.

This is true even to the point of deciding whether to promote your site with a www prefix or not. Many web sites can be reached both ways. If your site is like this, pick one format (either one is fine), and stick with it as your primary domain.

There are other issues that come into play when you have multiple domain names, so let's look at a fictional example. Congratulations, you now own all these domain names:

Assume that you've had the first domain for ages -- in fact, your site is called This is the domain you submitted to Yahoo way back in the 90s (wow -- we can say that now!), and it's the main domain you've been promoting when doing link building. This is the domain you want to continue promoting.

Assume the next two domains you registered last year. You did this thinking people might type them into their browsers, guessing that there might be sites at these addresses, just as people will guess that The Gap is at Fine -- resolve these names back to your web site. This lets you tap into that "accidental" traffic I mentioned. But don't link build using these domains, as that can dilute the value of your primary domain. Keep them "private."

Also, don't actively submit these domains to crawler-based search engines, in hopes of making them think that your single site is actually three different sites. Likewise, don't submit these domains to human-powered search engines, in hopes of tricking them into giving you multiple listings. Both actions could cause you to be considered a spammer.

Now let's assume the last five domain names were all just registered -- you grabbed the new long domain and all the hyphenated versions thinking these would help with search engines. But if you actively submit them to the search engines, you could be considered a spammer, right? Right. So here are your hard choices....

1) Submit some or all of them anyway to crawler-based services, and hope that you either don't get caught, that a competitor doesn't complain about you, or that if someone does complain, the search engines takes no action.

2) Create new web sites for some or all of the domains, then submit these new sites to the crawler-based services. You'll almost certainly see a traffic boost, though not because you have keyword-based domain names. Instead, it will be because by running multiple web sites, you'll have multiple root pages, which tend to be ranked more highly by search engines. You will also have taken a much larger step toward being labeled a spammer, assuming you are detected.

3) Don't actively submit the domain names, and just let them resolve to your site to capture accidental traffic.

Now even if you are absolutely conservative and never submit your "private" names as described in option 3, you may find that the crawler-based search engines will index you under them anyway. That because some scan for new names as part of their index building process. Also, if anyone submits your site using one of your alternative names, the search engines then becomes aware of it under that name. That means you could be accidentally spamming the search engines, even though you are only intending to legitimately have multiple domains to tap into accidental browser-based traffic.

Don't be too worried about this. If you haven't been actively submitting using the other domains, you probably won't have any problems. You should only take action if you suddenly discover your traffic from a particular search engine suddenly stops for no apparent reason. If resubmitting doesn't correct the situation, then get in contact with the search engine and ask if there's a problem with your site. If you've been placed on some type of blacklist, you should be told -- and you should be removed once you explain that you did not intend for the other domain names to be registered.

By the way, multiple domain names are a big problem for the crawler-based services -- they need to come up with better ways of ensuring sites aren't accidentally spamming them. And with this sudden explosion of new domain names being promoted as search engine secret weapons, it's likely they'll be taking a closer look at the issue. It's another reason not to bank too much on a keyword-rich domain name somehow propelling you to the top of the results.

ABCs and URLs

More information about the issue of keywords in URLs. It also discusses how keywords in domain names can have an impact in increasing traffic from within web browsers.

Password Finder

Registrars race to profit from longer domain names, Dec. 20, 1999

A look at the hype over long domain names, from a marketing perspective.

ICANN List of Accredited and Accreditation-Qualified Registrars

A list of all official registrars that can register domain names, though not all can yet issue the longer ones. And ICANN really deserves a ding for failing to post any information about the longer domain names on its site. This was a huge change, and as the body overseeing the domain name process, it failed and is still failing to keep the Internet community properly informed.


An accredited registrar that can handle long domain name requests -- no particular recommendation, just one that I know of offhand.

Network Solutions

Yes, you can also register long domain names with the granddaddy of registrars -- though unbelievably, it took the company nearly a month to catch up with its competitors in offering this ability.


Quickly discover if a domain name is available, along with whether a trademark for the term has been registered, discover related words, and more. A great tool. It's not yet long domain name capable, however.

Domain Notes

A great resource for all things domain. Helpful links, news and a wizard to find domain names.

What You Must Know About Long Domain Names
ActiveMarketplace, Dec. 1999

Want the hype? Here's a nice recap of various people saying that you should register long domain names to benefit with search engines.

Prank or profit? Web name auctioned off for $10 mln
Reuters, Jan. 3, 2000 may have been sold for $10 million -- now that's hype, if it pans out.

Search Engine Resources

KidsClick Worlds of Web Searching

Nicely done guide designed to teach children 8 through 12 the basics of searching the web.


Search Engine Red Lipstick

Still need a belated gift for someone into search engines? How about "Search Engine Red" lipstick, one of a range of Internet-themed lipcolors from Lipservice


Infonortics Search Engines Meeting

The 5th annual search engines conference backed by Infonortics has been announced for April 10 and 11 in Boston. It features speakers from major search engines and information retrieval companies, and it's an ideal conference for professional researchers, those with intranet search responsibilities or anyone interested in the technological end of search and retrieval. I'll also be leading off the first day with a review of search engine developments over the past year.

Search Engine Articles

Reconnaissance Over the Web
Washington Post, Jan. 3, 2000

Review of a new military search engine and searching for military information, in general.


AltaVista Search Engine Directory Traversal Vulnerability
BugTraq, Dec. 29, 1999

Describes a recently discovered security flaw in of AltaVista Search Intranet versions 2.0b and 2.3A. AltaVista has posted a patch -- click on the "solution" tab at BugTraq above or go directly to the patch via the URL below:

AltaVista Search Intranet Download Page


Portal envy doesn't always pay, Dec. 27, 1999

What to expect in 2000? Only two or three major portal sites, the analysts say. Of course, they say this every year, and every year they are wrong. Watch for 2000 to be ending with more major search and portal sites, not fewer -- as has been the case every year since these predictions began in 1996.


Web search results still have human touch, Dec. 27, 1999

Recap of major developments over the year in search.


Direct Hit files for IPO, Dec. 23, 1999

Direct Hit is ready to cash in on its popularity with a public offering. Some interesting facts from Direct Hit's filing. About 70% of its revenue currently comes from Lycos (which owns Direct Hit's flagship customer, HotBot). Former AltaVista chief technical officer Louis Monier sits on its Board of Advisors, which provides Direct Hit with feedback and guidance about technology and marketing. And, Direct Hit's relatively new Vice President of Engineering held that same position at Lycos until this past October.


Ask Jeeves denies infringing MIT professors' patents
Reuters, Dec.17, 1999

Two MIT professors claim in a lawsuit that Ask Jeeves should have asked them before infringing on their natural language search patents. Ask Jeeves says the case is without merit.


Altavista Free Internet Access Shatters Million Member Mark
AltaVista, Dec. 13, 1999

AltaVista says its free Internet access service has now surpassed 1 million members, after operating for only four months.


Netscape Directory Making a Splash
Industry Standard, Dec. 14, 1999,1151,8187,00.html

A look at the growth of the Open Directory, with pros and cons of the service.


New World of Web Reviews
Internet World, Dec. 1, 1999

A look at the emergence of comparison shopping and product review sites on the web.


How do I unsubscribe?
+ Follow the instructions at the very end of this email.

How do I subscribe?
The Search Engine Update is only available to paid subscribers of the Search Engine Watch web site. If you are not a subscriber and somehow are receiving a copy of the newsletter, learn how to subscribe at:

How do I see past issues?
Follow the links at:

Is there an HTML version?
Yes, but not via email. View it online at:

How do I change my address?
+ Send a message to

I need human help with my subscription!
+ Send a message to DO NOT send messages regarding list management or site subscription issues to Danny Sullivan. He does not deal with these directly.

I have feedback about an article!
+ I'd love to hear it. Use the form at

This newsletter is Copyright (c) corp., 2000