TR: How do you rank your pages?
TP: We developed something called Pulse Rank. We very much believe in the need to rank things. You've found something you want to learn about, but there are thousands of things about it, so rather than provide a real-time stream of things that aren't relevant. We rank the results in real-time, based on pulse ranking which broadly reflects current social buzz around content.
So, if the video is being shared a lot on Twitter and our panel are watching the same video a lot, then this will be ranked higher.
So, it might not actually be the freshest thing, but it should be the most relevant.
TR: How do you get on the panel?
TP: It's more like an internet measuring service. So you join the panel and then as you surf the web, you can then pass that information back to us anonymously if you so wish, and we can use that information to find out what is hot and what is not.
The way that traditional search works is through indexing and ranking. So the way that Google indexes is that it links one piece of content to the other and this spiders around the web.
In terms of ranking, page rank is basically heavily based on citation, so the number of links to a certain piece of content. So, if you have written an article and lots of other people have linked to it, then that page increases in page-ranking and so on.
As a consequence, this tends to favour the well-established websites. So, if you were to Google Iran, the first hit would probably be Wikipedia and the second page maybe the CIA page on Iran, as these will be well-linked. it's a dependable system but doesn't work with real-time search as well.
So, we at oneRiot sat there and thought how we could integrate what's happening now into this system. In order to do this, we had to invent a new way to index the web and also to rank the content.
The indexing comes from what it happening on the social web right now, and the ranking is what's the most socially relevant content within this – what is the must-see video to watch.
TR: Does this mean that your system sometimes throws up some obscure blogs?
TP: That's one hundred per cent correct. It's like a meritocracy – if you write the blog that everyone is reading at the moment, then you will be on top. But we have to be careful that what is popular isn't spam, so we also look at who is sharing these links and if they are to be trusted.
Working with real-time means that things are constantly evolving, and so the search results on our sites are constantly evolving as well. That's why a lot of people who use our site use it multiple times in a day as they want to be kept as up to date as possible and stay on top of trends.
From a business aspect this is great for us, as because our users search many more times in a day with the same query, we have the potential to monetise that user many more times during the day.
TR: What do you think of Facebook buying FriendFeed?
TP: It's going to be interesting to see what happens with this. There is potential there, so if I want to watch a video that one of my friends posted a while back, then I'll be able to search and find it.
Until Facebook opens up the search though, it's going to be quite limited. If, say, I want to know a good coffee shop in London, I can search around my friends group to see if they have mentioned one before but the chances of this are unlikely.
There are definitely going to be some useful stuff but real-time search – tell me what my friends are thinking about this – you probably don't care, you would rather read the New York Times or something.
Having said that, with 220 million users, if they could turbo charge that content and open up all that content like Twitter has done, then that will become a really interesting core of information.
For more on the OneRiot search engine, www.oneriot.com. Or you can check out the company's new RiotFeeds – a micro-site powered by OneRiot that offers an easy way for tweeters to discover the hottest links in their Twitter stream for any topic- tech news, gadgets, sports, green living, food & wine. Check this out at http://feeds.oneriot.com.
