New search engine Topsy, which has been in stealth development for three years, launches, well, now.
Before Google, search engines like AltaVista determined relevance based on how well a web page matched the query. Then came Google, which views the web as a network of documents. Today, all search engines analyze linking behavior around the web. When a web page is linked to a lot, it’s given more influence than other pages competing for attention around the same topics/keywords. Jeff Jarvis summed it all up nicely in 2005 “In this new world, links are currency. Links grant authority. Links build branding. Links equal value.” There’s lots more to it, but the notion that links create value is what drives Internet search.
Well, it’s no longer 2005. Back then blogs were giving Google fits because of how fast and irregularly they updated. Google had to make decisions on how often to index pages. Indexing is expensive, so there’s a tradeoff. Ping servers and blog search engines rose briefly to fill the niche, but Google indexes most popular blogs so often that those blog search engines are no longer much better.
Now, though, we have so much real time content being created that Google and the other engines can’t keep up. Most of this content is on Twitter, but FriendFeed, Facebook, Digg and lots of other services are adding to it, too. The result – more and more people are doing searches on Twitter Search in addition to Google. For me, someone who’s obsessed with news and stuff that’s happening right now, Twitter search is about 25% of my total Internet searches. The ratio keeps going up over time.
That’s where Topsy comes in. It’s not strictly speaking a real time search engine like Scoopler, which we wrote about earlier this month. Topsy is just a search engine. That has a fundamentally new way of finding good results: Twitter users.
The 30 million or so Twitter users are an army of little content-finding machines. Topsy says those users are sending tens of thousands of unique links per day to interesting things around the Internet.
Some of those users have more influence than others. And some links are sent by lots of Twitter users, others just sent once. Those links, combined with the information in the Twitter message itself, is what Topsy uses as the basis of its search engine.
And the results are…amazing.
New stuff in particular percolates up very quickly. A search for Facebook, for example, shows lots of news about the funding that was announced earlier today. And the links are sorted by those that Twitter users are sending around the most, weighted in favor of links sent by more influential Twitter users. You can sort results over all time (going back to September 2008), last month, week, day or hour. For all time, top results for Facebook are the Facebook site and developer site, among others. But in the last hour and day, it’s all about the funding news.
Results show popular links but also the most influential users tweeting about that topic. Click on that user and you’ll see all their tweets about the topic. Here’s the results for TechCrunch and Facebook, for example.
User influence is a hot topic, of course. Topsy isn’t looking at the number of followers. Rather, Influence is gained when others retweet links you’ve sent out. And when you retweet others, you lose a little Influence. So the more people retweet you, the more Influence you gain. So, yes, retweets are the new currency on the web. Told you.
Topsy was founded in 2006 and has raised nearly $15 million to date in venture and debt funding. More information on the funding and founders is on the CrunchBase page for Topsy.
Here’s a video where the Topsy founders give an introduction to the service and how it works:
Topsy sounds great need to check it out. Hope its not something like Cuil ;)
small tweak of ranking algorithm at google…
Lots of innovation happening these days in real time … Worlfram, Oneriot, Yauba, and now Topsy and soon the new Microsoft.
It would be nice to seen Google lose some market share for a change
Great work by a top group of Indian engineers … You make us all very proud! :)
Rainier said…
small tweak of ranking algorithm at google…
Yeah, I agree. Well doubt to the Topsy team, but I doubt they have invented something completely new, ie, original invention that has never been done before.
Anyway, to Topsy team, you may find some papers here from research scientists at Sandia National Lab useful for your product development. Look for papers on tensors also covered under higher order multilinear algebra.
correction:
I meant to say: Well done to the Topsy team…
and not : Well doubt to the Topsy team…
Real-time search is hot now and it’s definitely the next big thing to happen in search. One of my favorites in this space is http://www.boilingpage.com. The quality of its search results are simply great than any other real-time search engines.
eh not that big of a deal
15 million funding for topsy?
way more than needed.
i say 1.2 max
Even /that/ sounds like too much to me. In fact, for a basic, bare-bones search engine that does basically nothing but leech of Twitter and other social sites, the only funding I can possibly see it needing is among the likes of a few servers. I say a few because, well, let’s face it, look at the popularity of other “compile” search engines. The most popular one (last time I checked) was DogPile.com and, be honest, when was the last time you used it? It seems like it will die slow and painfully like Cuil, and every new “Search Innovator” that’s been over-hyped in the past year or so.
very helpful!
Looks pretty nice. Solid idea. Think Twitter will want to buy them? Twitter search needs much improvement.
Another silly idea…type in Marketing, for example, and find an article from the WSJ that is 3 months old. Is this supposed to be the “real-time” web? And where does this influence thing come in? It seems based on how many people’s tweets are retweeted – but that is a simple concept that will be gamed by spammers inside of a few months.
Are 3-month-old articles supposed to disappear from everyone’s conciousness? Maybe if you wanted something more recent, you could’ve actually clicked on the “Hour” selection instead of looking at the All-time results?
Not surprised that I’m having to explain this to someone whose first instinct was to search for “marketing”.
It looks interesting. Might be worth keeping an eye on.
I still think good content doesn’t always equate to popular content. If I use the WP widget for Most Popular Posts (or whatever it’s called), then people going to my blog will usually click on what’s popular, therefore making it more popular while the rest of my content sits there…whether it is worthy of being read or not.
Wow- Tinychat.com is #1 for “link”
..thats sort of weird, as I expected that to be the #1 most common phrase in tweets ( look @ this link, check the link, etc..)
http://www.tops...m/search?q=link
It is fast and has depth. The UI is difficult to scan. I think that the future of twitter search might be in hyper vertical micro sites. I like for example what Twazzup is doing around Google IO.
For example, compare:
http://googleio.twazzup.com
http://www.tops...arch?q=googleio
Hey mike
Topsy is actually really really cool. Nice way of finding and discovering content and oh so fast.
So they started it before Twitter started…. i wonder what the original plan was?
Not bad. Of the last 10 posts 4 of them have something to do or related to twitter.
Do you guys think thats too much?
Leave your comments below.
MIke, you started a good conversation on friendfeed about this yesterday: http://friendfe...erious-currency There’s some fascinating thoughts there.
thanks for that. changed the link at the end of the post to friendfeed, great discussion there.
This search engine sucks.
what do people think about http://twillage.com? It finds local events on twitter and clusters them by date.
Just a minority but there are people like @markismusing who does nothing but post quotes all day long and follows and un-follows people so his #’s rise. He gets a LOT of RT’s b/c we all know that people on Twitter LOVE quotes.
To me, he adds nothing to any conversation by doing nothing but copying quotes from a quote site, yet gets tons of RT value. So again, depends on what one is considering “value” in a conversation.
I’m getting extremely tired about realtime. I thought the internet was more of a get it when you want it thing, not a you have to be there one. It’s like tv…but I guess it archives.
What happens when Twitter is the next myspace?
Three thoughts …
1) Wow, this is the tool that twitter needed to build to monetize their platform.
2) My guess is that a few people at Google are having a meeting right now too.
3) Marketeers are already working o engines to mass re-tweet stuff.
oh, forgot one…
4) Don’t forget to re-tweet this http://twitter....atus/1931135703
An Another Twitter Article… I like Twitter but frankly – I must agree with some of the previous posts. It’s too much.
I suggest you to setup a new blog: TwitterCrunch.com – what do you think?
interesting…..let me check
Wow…. it tooks three years to build this?
Utterly convenient timing to if you ask me
what they did with $15m in 3 years, I am going to do in 10hrs with a better UI and more powerful engine… wait and watch.. what a way to waste VC money. Good job to the Topsy team.
Not sure where they needed that 15M? Looks like they work on a cheap table and metal chairs. Either very conservative with cash or very stupid….
sad to say they spent 3 years doing nothing and then decided to go with the media darling that is twitter. You cant tell me they have been working with the twitter idea for three years.
I think Google index’s things well enough. The key to beating Google is having an algorithm thats better than Pakerank. Links are no longer valuable measure of quality .
With 15 mil and three years im sure I could buy the brains to something that performs better than Google. Take a look at what Matt Wells of Gigablast has achieved with little money.
Guy Kawasaki is gonna have a field day with this. The search results aren’t bad, but regardless of whether or not it’s good, Google owns search mindshare and that wont change because some company tries to staple itself to an overhyped micro-blogging service.
Spamming this engine is going to be hell of a lot easier if in-case this becomes popular.
Wondering if anyone is working on something that can offer info on the various keywords being searched in these engines and twitter searches.. something similar to what wordtracker does for main stream keyword searches..
Realtime search is evolving so fast, this is an interim step imho. Why not search @, or url’s, or #, or some combination for that matter, its unproven that RT’s drive high search value.
And as someone noted, the minute anyone realises that search will be optimised on RT, spam and SEO will begin bigtime….
RTs are not a good measure of authority, they are too easy to game, plain and simple. Something like TunkRank is better, but one can easily show (via simulation) that with a small number of bots RT can be easily gamed. Also, do a search for Barack Obama, you’ll see how authority is skewed, just because someone gets retweets w/ Barack Obama in them doesn’t make the tweet, or the link that was tweeted worthwhile. One needs to to analyze the actual CONTENT of the link and tweet to determine that.
My site Re-searchr.com does a better job of analyzing actual content, and I have spent nothing but $2K and sweat on it… 3 years and $15M unbelievable.
DecisionSearchEngine.com
Internet is now a vast garbage. The more people, more and more garbage … until the end of the world.