TechFold - Bold tech & web commentary

Bold tech & web commentary

TechFold is technology discussion, commentary, reviews, and opinions from well outside the valley. There's no koolaid to drink here, and TechFold is not in SL, or on Twitter.

Predictions for 2008, #3: Semantic apps will continue to suck

Well, “suck” is probably too strong a work. How about “Semantic apps will continue to be completely irrelevant?”

2007 was a year heavy on buzz around so-called next generation, 3.0, semantic apps - things like PowerSet & Hakia (natural language search), Spock (people search), Twine (who knows), and so on.

Well, after a year of anticipation, each is either unlaunched (powerset), unloved (spock is weak), or unheard of (twine).

Consider the value propositions:

Spock promised powerful people search, and so far has delivered a (drum roll)… social network scraper. See “idiocy writ large” for reference.
Next gen search engines promised better results than google and so far have delivered… junk. See my off the cuff Hakia vs. Google comparo for reference.
Apps like Twine promise to automate the discovery of relationships between your disparate bits of information - but if you listen to the video there’s a lot of buzzwords (semantic graph! open! graph! wikis! ontologies!) and little substance (its got smarts!) beyond keyword parsing.

…so my prediction is that Google will continue to kick ass, the so-called next-gen startups will continue to languish or sit or beta-purgatory, and slowly the concept of “semantic” will fade from the public eye as VC’s find something else to latch onto.

(Quick update: Twine has also committed what I consider to be the cardinal sin of marketing plans - attempting to create their own vernacular, i.e.: in scoble’s video above one “twines” a link, and creates a “twine” about something - arggh. Its as bad as the now defunct Teqlo)

2008, google, hakia, powerset, predictions, semantic, spock, twine web3.0

Hakia vs. Google: the 5 Million Dollar Question

Hakia just loaded up with another $5M to keep working on their semantic search “web 3.0″ product.

Out of interest, I compared search results for what I thought would be a good “semantic” search on both ol’ faithful Google and newfangled Hakia.

The query was: “How to protect my privacy on facebook.” Hakia’s results are here, Google’s here. Screenshots and commentary follow.

I figured it would be a good test of the two because there’s meaning to be extracted from the semantics of the sentence - “how to” implies that I’m seeking a tutorial, “protect” means that I have a specific action in mind, as opposed to just seeking editorial content on “facebook” and “privacy” and so on.

The verdict? Hakia’s second result was a splog trying to get me to sign up for a credit assessment. The rest of the Hakia results were a low quality mish-mash of blog posts and miscellaneous content which by and large were discussions about Facebook - for example, a bunch of HuffPo stuff.

Google on the other hand, started with a PDF from Canada’s Privacy Commissioner on how to adjust privacy settings in your Facebook profile, and continued with several highly relevant tutorials (including Facebook’s own privacy pages), before breaking down into discussion posts.

The bottom line? Google’s result was massively superior. On the other hand, Hakia has built in social networking, offering to introduce me to others who have received similarly dissatisfying results for the same search. Super! Ok, fine. I’m not giving Hakia a fair shake with a sample size of one. But nor will anyone else, and first impressions count. At least Powerset keeps their weak beta results private; Hakia should have done the same, IMHO.