Stumpedia Stalled?

A couple of weeks ago I posted about Stumpedia - a recent entry into the growing ranks of Human Powered Search engines. At the time, I wondered aloud about Stumpedia's value proposition. While honestly believe that real live people are the solution to the search relevancy issues we face today, Stumpedia seemed caught in a classic Catch 22. They're not crawling and rely on user link submission to build their index, but since the index is so sparse the results are poor (mostly non-existent) and thus there's little reason for anyone to adopt Stumpedia as a search service. But without users, how will they get links? You see where this is going.

Anyway, at the time I decided to start tracking the following (as reported on the Stumpedia home page):

• Registered Users
• Links
• Search Terms

A quick (and unattractive) excel graph shows the growth trend in all of these beginning to flat-line.

In addition to these slow growth trends, their company blog and forums have gone quiet. No posts to either in nearly two weeks. I hope they're off somewhere figuring out how they're going to bulk up their index, or they could be road-kill in the near future. I'll keep watching and report back.

Rushmore Drive

The Wall Street Journal is running a story on IAC's start-up strategy. Included in these is RushmoreDrive.com - a vertical search engine targeting the African American community. What's most interesting about RushmoreDrive is its demographic approach to a vertical search strategy as opposed to a content-centric approach. As such, it's easy to see how RushmoreDrive could easily evolve into a broader social media site - and you can even see social interaction elements on the site already.

Out of the box, it's nice to see RushmoreDrive allowing users to rank search results. As with any search engine, how they determine relevancy is a critical factor. For anybody who knows me, you know I'm an avid proponent of abandoning link-based page rank in favor of user response to search results. RushmoreDrive is using the Ask engine, but identical searches on the two sites yield different results pages. For example, a search for 'Baseball' on RushmoreDrive yields the following top 5:

The search on Ask.com returns Baseball-almanac.com, MLB.com, Baseball-reference.com, sabr.org (The Society for American Baseball Research), and baseball-links.com - in that order. Clearly RushmoreDrive is able to alter the default relevancy rankings returned by the Ask engine, and at first glance this seems to make sense. That is, for a search engine that markets itself to the African American community, prioritizing black history in a baseball query appears to be logical. What I'm unclear about is how well this type of search result reflects the user intent.

In my mind, a search for 'Baseball' is a generic search - no matter where it's submitted.  If RushmoreDrive were attempting to curate results and return only those believed to be of interest to its user base, that would make sense. But in this scenario, the same results returned by Ask.com are available, just pushed down the list a bit.

I'm on the fence on this one. Hopefully as the site gains some traction, the voting mechanism will catch on and the users will influence relevancy. It will be interesting to check back in a few months to see how these rankings may change.

Stumpedia

On February 22, 2008, Stumpedia launched as another entry into the Human Powered Search Engine category. Stumpedia differs from Mahalo, however. Rather than paying staff to generate targeted search results pages, Stumpedia depends on the good will of the community to submit links and suggest relevant search terms for each.

SearchRank.com has published a nice overview of the Stumpedia service if you want to get acquainted with exactly how it works. In the end, it's very easy to use and intuitive to navigate. After playing with it for a while, here are some observations -

Human Powered Search
It's pretty clear to me that the holy grail of search lies beyond the algorithm. Any attempt at human powered search seems seems at first glance to be untenable. The sheer volume of information is overwhelming and it's daunting to consider the number of people required to perform the task. Mahalo's approach to employ staff around the country (globe?) to build hand-crafted search pages seems unmanageable in the long run, especially when you consider the rate at which new web content is produced. It's also hard to consider their pages 100% objective when they're produced by paid staff.

The potentially better solution is the path Stumpedia has taken - let the searchers themselves indicate which sites are relevant for given search terms. The risk in this model, that Mahalo avoids, is abuse. If Stumpedia starts to get any traction it will become a magnet for online marketers, spammers, and the like. At that point the quality of the search results become suspect unless Stumpedia can somehow keep the hordes at bay. Their stance on this is that they expect the community to be self-policing. How well this will work remains to be seen. They shouldn't lose site of the fact that a non-trivial portion of their user base will be marketers attempting to influence the results pages - and these folks tend to work together more often than rat each other out.

Google Indexed Search Results
I covered this point when looking at whether Mahalo fits Google's definition spam, but it's curious to see the Stumpedia search results pages being indexed by Google. For Stumpedia to be genuine in their attempt to become an alternative search engine, they should block the spiders from indexing their results pages. One search engine garnering traffic from another is simply a terrible user experience. Stumpedia should be able to stand on their own merits and attract a user base through the quality of their results.

Value Proposition
Stumpedia is in a bit of tricky situation at the moment. Since they rely on site visitors to submit links, they're going to depend heavily on content producers to join up and start adding their sites. Publishers will only do this, however, if there is a large enough user base such that they can expect a reasonable amount of return traffic. But currently there simply aren't enough links submitted to the site to make it useful for the searcher and so I expect adoption will be slow. So we have the publishers waiting on the searchers and the searchers waiting on the publishers. Stumpedia will likely need to lower the barrier for link submission (RSS?) to encourage publishers to jump on board. Right now, link submission is manual and I doubt major publishers will be keen to dedicate man-hours to submitting links individually.

There are some quirks I noticed while using Stumpedia, but I expect these will get flushed out as the service matures. I've submitted a few links as a test, to see what kind of traffic it generates. Stumpedia also displays the number of users, submitted links, and search terms directly on their home page. I've started to record those numbers and will check back a few times a week to track their progress.

Related Posts: Stumpedia Stalled?

Yahoo Semantic Web Adoption

SES NY closed this past Thursday, leading off the last day with a presentation by Yahoo Chief Scientist Andrew Tompkins. Mr. Tompkins' presentation centers around Yahoo's recent announcement that it is embracing a number of  Semantic Web standards.

As presented by Mr. Tompkins, Yahoo is attempting to shift the focus away from the 10 blue links and aim it's sights on user intent. That is, users are conducting a web search to accomplish a specific task. Whether that is task is to find a good Chili Recipe, get the phone number of the local hardware store, or find the top rated steak house in New York City, these tasks center around specific content consumption. Yahoo hopes to bring that specific information to the surface as quickly as possible, thus satisfying the user intent in the most efficient means possible.

Several examples were given in the presentation and Amit Kumar, Product Management Director for Yahoo! Search, gives a LinkedIn sample on his blog.  For a user search for him, rather than just returning a link back to his LinkedIn profile, the search will see something like this:

You can image similar search results that can surface restaurant rantings, contact information, movies times, and other specific details about the search subject.  To accomplish this, Yahoo!is supporting a few different mechanisms that will allow site owners to send specific page meta data  (i.e. the semantics) to the search engine.

The first of these is the Resource Description Framework (RDF). Using RDF, site publishers can embed meta content tags directly into each web page. These tags would describe the key content themes of the page, which the search engines can then pick up and subsequently use to inform search result ranking and presentation. To achieve similar results, Yahoo! is supporting Microformats as well as several semantic web vocabularies  (Dublin Core, Creative Commons, FOAF, GeoRSS, MediaRSS, and others).

Along with Google Universal Search, this is another example of a major search engine acting more and more like a destination. Yahoo! already brings much of it's own content to the top of the search results. Yahoo! Local content headlines a search for "NYC Steak House"...

... and Yahoo! Food gets top billing for "Turkey Recipes."

You can also now play YouTube videos directly within Yahoo! search results. From the search engine perspective, keeping the user on the search results page longer is preferable. They don't don't monetize those clicks from the organic results, and the longer they hang around the more likely it is they'll eventually click on a CPC ad.

From an SEO perspective, the rules of the game are changing. As the search engines become the information source, organic clicks will decrease. Bucking the trend won't be an option since you can expect sites that open themselves up and provide the rich meta data the engines want will surely find their way to the top.

For some businesses this will work out well. Local businesses should be happy to have phone numbers, addresses, maps, etc. exposed directly on the search engine result pages. After all, they're online to be found and mostly get people into the store / bookstore / bistro. Content producers, however, and other sites whose monetization strategy is largely tied to page views will face a challenge. If the latest scandalous picture or funny video or restaurant rating is delivered to the user without them ever having to leave the search engine, organic traffic to these sites will surely suffer.

Mahalo

I was impressed by the Jason Calacanis keynote at SES NY today. Mr. Calacanis is a serial entreprenuer, current CEO of Mahalo, and has been the center of a fair amount of SEO controversy.

What impressed me most about Mr. Calacanis was his conviction that search needs to evolve and, specifically, algorithmic search is not the right answer for every query. This thinking is the rationale behind Mahalo - a human assisted search engine. Every result page on Mahalo is created by a person, and every request for site inclusion is reviewed by a person. By using people to determine site relevancy for a given search term, Mahalo and Mr. Calacanis have given us a true measure of relevancy and effectively taken away the lion's share of what the average SEO bills her clients for.

Mr. Calacanis also commented that he believed SEOs today are too focused on the near-term win as opposed to a longer term strategy of building value. In defense of SEOs everywhere, they're only partially to blame for this. The link-based page rank algorithms promote a culture of deception. It's either too tempting to try to game the system or there's no choice becuase all your competitors are doing it. How can Mr. White Hat win when all those shades of grey exist and his clients are being promised hundreds (or thousands) of inbound links?

Perhaps this is why te SEO industry has been so defensive when it comes to remarks Mr. Calacanis has made in the past. If SEOs were secure in their craft, why would they care what anybody says? The writing is on the wall and the SEO industry sees the changes that are coming. Universal search, hand edited page rank, human assisted search engines, and even social media threaten to alter the fabric of SEO, and the situation has more than a few people scared.   

Rather than claiming every blog post which questions the current SEO industry is "SEO Hate," it would be better to embrace the fact that search, and thus SEO, are immature industries. Spurred on by advances in technology, the search industry will continue to evolve at a rapid pace and what we consider SEO today will be vastly different just a few years from now - and those that continue to focus on the short term will be left behind rather quickly.