Lavva: A New Attempt at "Social Search"
After seeing how hard it is to combat the goliath that is Google when it comes to search, you almost have to wonder about anyone launching an alternative search engine these days. Are they crazy? Overly ambitious? Probably a little of both. The latest attempt to snag a little search market share comes from Lavva, a company with big ideas about social search. Instead of retrieving sites based on a search algorithm like Google does, Lavva bases its search results on what people say are the top results. According to the company, this makes search "100% democratic." After a few test searches of our own, we can only say this: there's a reason why Google is king. Algorithms work.
The "Social Algorithm"
In theory, the idealism which infuses Lavva sounds like a good idea. "Search powered by the people," "results based on quality, not SEO," "transparency," etc. are the types of things Lavva likes to rave about when discussing their social algorithm. In practice, however, social search like this doesn't work. Obviously, it doesn't help Lavva's case that very few people know their startup even exists. Without users to rank the results, there's just no way Lavva can highlight the quality content.
Their idea just misses the mark, unfortunately. They would probably have had better luck if they overlaid their social algorithm on top of Google results, for example. That way the most relevant links would be retrieved first and then users could rank the results based on quality. (Rankings are done using little thumbs up and thumbs down icons beneath each result).
Do You Want to Chat about the News in Your Search Engine?
Another one of Lavva's hair-brained schemes is their "News Goes Social" page. Here, the engine aggregates top stories from a select few resources (CNN, Reuters, BBC, UN News, AP) and combines those with top search terms and the top links on Twitter.
While this in and of itself isn't entirely crazy (or entirely useful for that matter), how they want you to interact with the content sort of is. News stories have a "go social" link next to them which takes you to another page where you can chat, debate, and discuss the topic with other online searchers. After clicking through a number of these links, it was clear that no one was using this feature.
In a similar vein, users can sign into Lavva and click on the comment bubble icon under search results to leave their comments on the news story. Remember when Google tried this? Yeah, it was universally disliked then too. Frankly, this just isn't how people want to interact with search. And if Google couldn't make voting and commenting on search work, what hope does Lavva have?
Future Plans: Twitter and Facebook
While Lavva's service may get a little more interesting when they release their next update which plans to incorporate search results from sites like Twitter and Facebook, we doubt that alone will be enough for Lavva to make any impact. Even a startup as promising and innovative as the social search service that was Delver didn't make it, eventually selling out to Sears (yes, Sears!) in the end.
For now the best thing that can be said about Lavva is that it's powered by hydroelectricity, making it one of the greenest engines around. They plan to move to a solar-based system in the future, reports Seattle tech blog TechFlash. While we're happy that they're concerned about the environment, going green isn't going to be enough to make this social search attempt work.
Share
Facebook
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Very true. There's a reason Google wins. Data. Lavva's whole concept is years late even to be trendy, much less plausible.
But it's a bit unfair to pick on a sample search to prove how bad it is in a short article like this. I have no doubt Lavva is bad, but you can find horrible search responses for any engine. To really understand how awful a search engine is, you have to run a whole bunch of queries.... Then you realize, "Yeah, Wolfram Alpha just sucks for most queries" or "Ask Jeeves? I thought Jeeves was supposed to be a wise and clever valet, not a retarded mynah bird with tourette's syndrome...."
But Lavva does win a small prize for most ambiguous pronunciation award (since the double v looks like a funky w in many fonts), stealing that coveted gold ring from Cuil.
@Miramon: Believe me, the example shown was just one of many, many, many searches I tried. Do some searches yourself and see what I mean.
I guess someones got to try...
As you rightly point out, search engines that rely on crowdsourcing to generate correct results require a crowd (defined as alot of people). That might not prove too difficult even in the short term for popular search terms - but it means less used search terms are doomed to give unreliable results - so as you say "algorithms work"! Consistently.
actually google does the voting of results as well. It might just be an experimental thing that certain people have access to or maybe when you're logged in, but if i could post a screen shot, theres a up arrow that you can promote links for you personal search results, and I imagine with votes, it must somehow be included in their algorithm.
Sorry Sarah - re-reading my previous comment, it seems I didn’t really add to what you had already said except to emphasis that they needed a crowd…
What I meant to say was that, in order to build a search engine on the basis of crowd sourcing, they need a pretty big daily crowd. Now despite my appalling history with mathematics, I think this has got to be worth a number crunch, because my initial guess is that but that crowd must probably be in the tens of millions on a daily basis for the following reasons.
In order for a representative sample to crowd source select the SERP’s for a specific keyword accurately, you must be talking hundreds of people. Let’s say 300.
Then there are going to be the people who search but don’t select – I seem to remember that you can count on much less than 10% to be do’ers. Say 10%.
So, so far, that means that for every 1 million visitors, you might be able to accurately crowd source SERP’s for 300 odd keywords.
According to Webster’s dictionary, there are approximately 460,000 words in the English language (other dictionaries say over 1 million with scientific words). However people don’t use anything like that generally – let’s guess at a general vocabulary of 1,000 words.
The problem is though that search terms (keywords) can be one, two or three words long in the majority of cases. That gives a spread of keywords that you must have answers for of 1,000 – 1 million – 1 billion keywords. Let’s say the majority of searches fall into those 1 million keywords.
So, in order to get “accurate” SERPs for all those 1 million keywords, on a one off basis, you would need 3 billion visitors. Spread that over a year and by the end of the year, with a million daily visitors, you would have your answers – except most of them would be out of date.
Google could probably get it to work, but a start-up?
As an aside, and to partly explain why the above topic interested me so much, Google has a great tool called the keyword tool, which can generate a huge amount of keyword variations for a given keyword. It also tells you the number of local and global monthly searches on a particular keyword combination. Or does it?
I’m CEO of a free online global address book. In fact we currently rank number one on Google.co.uk and number 6 on Google.com for the keyword “address book”. Pretty good so far. Lavva take note – that’s a two word search term, neither of which makes sense to our site as individual words.
However, Google says that that search term gets 135,000 searches in the UK per month and 673,000 searches monthly. Our site is seeing less than 1,000 monthly hits for that particular keyword. Needless to say, that falls hugely short of their keyword tool traffic calculations – even if you factor in that not everyone in the UK is using Google.co.uk (a Blogstorm article reporting on Hitwise divulged that Google.co.uk had 75% of the UK market and Google.com had 15% - back in January 2008). So where’s does Google get its traffic calculations from and why the apparent discrepancy?
There are plenty of social, web 2.0 and real time search engines that work well already like http://www.Yauba.com and http://www.Topsy.com
Yauba does not offer commenting, but its results use social algorithms as does Topsy.
Google doesn't deliver results; It delivers starting places to query. Search engines that use the Webcrawler method work best for research, assuming an article or a paper is crawled and instantly delivered.
Social search, in terms of social relevancy with respect to search terms, makes zero sense, and would take ten times the effort.
Social search in terms of socializing makes sense. You have to search to socialize.
The best way for people to get what they want is to get what they want directly from the source or supplier.
Read http://inversearch.blogspot.com
Here's a real-time social search engine I created and would welcome any feedback / constructive criticism on how I could make it more useful for folks.
http://www.searchmotive.com
Thanks!
I recently came across your blog and have been reading along. I thought I would leave my first comment. I don't know what to say except that I have enjoyed reading. Nice blog. I will keep visiting this blog very often.
Margaret
http://grantfoundation.net
very thanks for article
I love you more ideas. it seems like another blog I thought I would leave my first comment. I dont know what to say except that I have enjoyed reading. Nice blog. I will keep visiting this blog very often.