Social Search Sites Could Challenge Stalwarts

The following is also my column in this week's Advertising Age.

Search is broken. Of course, with Google at more than $600 a share, 91% of us using search engines and studies showing we're largely satisfied with their results, it's easy to discard this statement. The problem, however, is that search engines can bring back too much information. With content becoming a commodity, it can be difficult to separate the diamonds from the duds.

Thankfully, a number of smart entrepreneurs recognize this, and they are on the case. They see an opportunity to create a new, blended approach to search that allows us to scour the web just as we do now but with more guidance from community curators.

Mahalo, which means "thank you" in Hawaiian, is among the most notable of these upstarts. The site, which launched with a great deal of fanfare in May, is the brainchild of serial entrepreneur Jason Calacanis.

Mahalo blends wiki technology with search. The site has a small team of editors and even more volunteers who work to pull together frequently updated pages that point users to high-quality links for the top 10,000 searches in popular categories. These curated pages cover topics such as how-to articles, the latest gadget reviews and more. Pages are updated frequently as news breaks.

If a page does not exist in its database, Mahalo will aggregate results from all of the major search engines, including Google, Live.com and Ask, as well as Wikipedia and YouTube. Further, users can apply to become a guide or suggest pages and links. The only way to advertise is through contextual search ads placed through Google AdSense.

While anecdotal data shows that Mahalo may be getting some traction, it has a lot of competition in the same genre. Wikipedia founder Jimmy Wales is preparing his own social search engine, Wikia. It promises to be more open. About, which is owned by New York Times Co., has long taken a similar approach. Finally, Google too is showing signs of becoming more social. Just recently it started allowing users to edit maps or collaborate in the open to build complete travel guides.

Comments

[Deepak]

Steve,

85% of what I am interested in does not exist in Mahalo and most probably never will. The challege for companies like Mahalo is scale, and scale is Google's secret weapon. I prefer the approach taken by Lijit, which adds a trust layer onto Google (which is still, by far, the best engine out there). The algorithm first, with relevance enhanced by your social graph if that's the term people prefer (circle of trust is a better one) is the better approach. In fact, one could argue that your real social graph does not make search better. My blogroll on topics of interest does, since those people are authorities on the subject, but they might not be friends of mine on Facebook or anyplace else.

[Pete]

Hasn't delicious been doing this for years?
I rearely find rubbish results in it, compered to google.

[Andrew Goodman]

Not much traction here.

Looks like Mahalo is getting its butt kicked by Squidoo, at this point.

What does that tell you?

To be considered a "leading search engine" Mahalo will need to become a top 100 site. Right now, they're fighting to crack 1,000.

[David Scott Lewis]

Steve, Mahalo may do well with top 100 searches, but can't realistically scale, at least not the way they're doing it. An About.com approach may help, but this also seems too limited. However, they could outsource a lot of the editing to help keep their HR costs in check. In this fashion, perhaps they can realistically handle the top 1,000 or 2,000 searches. It would be interesting to see what percentage of searches are encompassed within the top 2,000 searches. There might very well be a good business model for Mahalo, especially with offshoring. (Have they opened a branch in Manila yet?)

A better approach will be with search results derived from contextual Semantic Web-based social networks. Twine is one -- and perhaps the best -- example. (Twine is now in private beta and I'm an active contributor in their beta.) Powerset will also have a social networking component. (I'm in their public beta, although at this point, the social networking component is overshadowed by their NLP capabilities.)

Frankly, I would argue that search results are better when confined to an edited database of selected items. I know that if I'm not searching for recent info, a search of my Furl archive will almost always give me better results than a Google search. This, of course, assumes that I'm looking for something that has a lot of items in Furl. del.icio.us is OK for this, but with Furl's digital library functionality (something del.icio.us doesn't have), Furl does a lot better at searching within one's own archive.

But there's still a problem. The problem is called "tagging." Tagging is stupid, a complete waste of time. At this moment, however, there isn't much of an alternative. Yet, think about Copernic Summarizer type "Concepts" (which are akin to tags) through automatic summarization. Now take this a step (or two or three) further with semantic processing and there's a whole new game in town. No need to tag: It's done for you ... and it's better than you, too. And someone can always add tags that are specific to their worldview.

At this point, social networks are formed based upon contextual proximity and one's sources of information also explode as a result. Searches are then done within one's expanded digital library and then get pretty deep, much deeper than the top 2,000 searches.

This is the way search AND social networks need to go. Not ALL search needs to be done this way. For a top celebrity, use Mahalo. And for some people, especially people who do very little reading or browsing, Facebook or MySpace may be a better option: Plaxo + a way to share one's life. But for a lot of people, Twine will be the best way to search AND the best way to form social networks.

[David Scott Lewis]

See http://doiop.com/Twine.

[Steven Krein]

Steve - good article...saw it in AdAge and glad you posted it here and are covering this trend....

The notion of a new "service" layer on top of the great technologies, like Google's, is only going to pick up over the next 12-18 months as human-powered search services like Mahalo and more vertical versions like OrganizedWisdom Health (http://organizedwisdom.com) build out high-quality libraries of tens of thousands of pre-published search results. These services help people who don't want to spend hours and hours searching for information in google, yahoo, ask and delicious, when it can much more easily be delivered to them on a silver platter.

I just wrote an article about this trend for CMP's new online publication Internet Evolution that is directly relevant - http://www.internetevolution.com/author.asp?section_id=557&doc_id=140592&

The fact of the matter is it's really really early. There is traction here...and eventually it actually will go beyond search. I guarantee a few years from now everyone will be amazed with how many new services will exist that help the masses do things online that today they have to do themselves.

[Paul Madden]

Steve,

Lots of I's in the comments, which is fair enough, but how mainstream users search and might use a top level human search engine like Mahalo is more interesting.

Clearly, Google is still growing and will be dominant for some time to come (perhaps for ever). Most new users will follow the trend of the collective, grab whats closest at hand and what they perceive to be easy.

Google is popular and easy in terms of requests, but not always in providing results. They do seem to be learning and making local search/maps more collaborative and this will help in cutting through the increasing mass of information. Consumers will target their searches better, review and comment on services. Data owners will work harder at making their content more relevant and searchable. But will it be enough to cut through?

Mahalo could have a future role in our searches. Somewhere to go if Google makes no sense or doesn't produce the results required. They may just focus on the top level searches, limited subjects, perhaps a few, but wide verticals, providing more relevance and depth of data in one place. Nothing wrong in that if it can provide a solid return.

I think results that are collected by a Human have appeal and value, its something that the average user can understand. There is also the two-way interaction, the ability to comment on results and comment on what is needed for the future.

Ultimately, Mahalo is easy to use, results make sense and the branding is simple and engaging. They do need to consider a more open model, but regulating the value of content is hard. Mahalo may attract the mainstream, more than something like Squidoo - will the average user get the concept of a lense?

[Ram]

The emerging power of local search i think is being underestimated. I just read article where the current local search stats do not take into account ambiguities in search by users and lack of geo modifiers, meaning the stats is more than just a billion searches a month.
Search engines have a clear role to play - present a macro view based on relevance, content depth, popularity etc which are basically qualifiers for higher placement( everything web).

Local search is more mundane and transactional(everyday web) and needs a hierarchical database to funnel the user through to locate.

Seach engine is like a plane which u dont use to reach ur office in the city but u use local transport- Nothing wrong. Nothing wrong about the aeroplane too.It is just not the vehicle for some purposes.

One size fits all will be the doom for all.

[Petterson]

Google's search results are based on mechanical analysis of web pages. Its algorism is a bit advanced more than other search engines. So it makes great success. But this advantage is not for ever. I bet there will be a more advanced algorism based on human understanding in the next one or two years. It will make bigger success than google does.

[Michael Lodispoto]

Hi Steve, I think that EarthFrisk.Org, which is released now but not officially until Monday is a better search as it is a hybrid Meta-Community that touches on a number of points you have made in this article and others. It has an interface that allows for fast searches of Digg, StumbleUpon and Wikipedia among other sites. More importantly rather than trying to beat google, it is Meta in that it has the best results from Google, Yahoo and MSN and others plus irt gives the community the opportunity to then vote and comment on these best results. The comments and votes translates directly to a new rating system for both websites and individuals called CV rating ( Color Value). It's complete with downloadable toolbars for Firefox and IE and since it is the first true community web 2.0 search engine with the best reults of the majors, I think we can outdo Wiki's release in another week or so.
The press release is set for Monday with much media and blog and more press. The basic theory is the people can do a far better job at preventing spam than any search engine and over the next weeks and months the results will evolve beyond Google and Yahoo and other major search engines. We've created our own patent pending technology that as of Monday will be completely transparent.
So we beat Wiki people to the punch and have far more features. The new rating system can indeed offer a serious challenge to Google's pagerank.
Only a few more things to work on before Monday and the coders and programmers (on 3 continents) are done with the Beta release.
The images and Video need to be made more beta but the search and most importantly the ability to comment and vote on results can help evolve search to the next level. A true web 2.0 search engine is about to be born! Hope you like it on Monday when released.