Social Relevancy may turn the web into a next edition of the Lonely Planet

(disclaimer: this post is written using an iPhone and a crappy Internet connection from a sunny beach in Greece. It is bound to contain (spelling) errors 😉 )

I just read a good post by Alex Iskold about social search. Pagerank is currently the most relevant search algorithm, both in terms of delivering information and revenues. Finding the next ranking system is as much a quest for new relevancy as it is a treasure hunt.
Social Relevancy Ranking is a direction that sounds promising but also raises a lot of questions for me:
1. How will relevancy be determined? Robert Scoble is a great source for tech news, but can I trust him when it comes down to movie reviews, cooking pasta, or the theory of black holes? How can I, unbiased, determine who is an authority, or can be trusted? This is hard to determine within a group of people that you really know, so how can you do this for strangers?
2. How can the algorithm provide unbiased results? If anything ‘social’ has yet again proven us that it is human to cheat and bypass the system. Just look up any discussion on Twitter’s SUL list and you will know what I mean by that. Algorithms are unbiased, people cheat.
3. How will signal be found in the noise? Pagerank takes time to build. You need consistency and quality. But how can we deal with that when real-time results are incorporated? What if results from bots and spammers start to dominate the input, as they seem to be doing on services like Twitter? How can we distinguish commercial play from an honest message (this is the best coffee I’ve ever had).
4. How can we ensure that the web will not become a next version of ‘the lonely planet’? This is my worst fear. The web and the emerging technologies make everything and everyone look the same. There is no room for individualization, uniqueness, or creativity. It will all be captured, indexed, socialized, and pretty soon we are all looking at the same web. We will all discover that great place that serves great coffee, the one authority on backpacking vacations, the best car to drive, movie to watch, recipe to bake. Social relevance ranking becomes the common denominator. It makes the world flat, and leaves little or no room for alternative voices (as they will not rank high).

Social relevancy ranking may become the next big thing after PageRank. It will be exciting to see where it will take us. But it could very well turn out to become the biggest threat to our most precious gift, our own individuality!

About vanelsas

See my about page, ;-)
This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

6 Responses to Social Relevancy may turn the web into a next edition of the Lonely Planet

  1. I have often wondered about the self-selective effect that can stem from users bumping content which is already spotlighted, as a result the long tail can get longer and farther away, less attention to non-spotlight content.

  2. CJ says:

    “Pagerank is currently the most relevant search algorithm, both in terms of delivering information and revenues.”

    That’s not true. PageRank is a link analysis algorithm, it is not the entire ranking method for Google. Lots of other variables enter the equations. Also the IR community is still unsure as to how to evaluate IR systems efficiently, you can read more about that in the TREC streams.

    I don’t think Google gives the best results every time. Sometimes yes, not always.

    “How will signal be found in the noise? Pagerank takes time to build. You need consistency and quality. ”

    I’m not sure what you mean, do you mean spam? Not all noise is spam of course. Do you mean a site takes time to build up PageRank over time? I don’t think that there is any consistency in unstructured search and quality is dubious. We’re not really sure what that is, and to be honest I think it changes with the context and user queries. Do you mean rather that the site has to be consistent and of quality? If so I agree with quality but not sure about consistency.

    Personally I don’t think that incorporating social results in the standard search results is a good idea. That social data is really different from all the other static data on the web (both structured and unstructured).

    I think that if you look at incorporating or even ranking social results with the same technology as for the rest of the web, then you are going to get what you refer to as a flat web, but worse still a right mess. Thing is that the algorithms are different for social real-time stuff and the rest, and there’s personalisation as well don’t forget.

    Also, PageRank whilst being super important in computing and having been a rather interesting breakthrough remains the “the most well known algorithm” for SEO’s and webmasters mostly. My mum has no idea what it is and my computing pals will fondly remember BM25 for example. What about GFolkRank, LETOR, and all the “learning to rank” research? Thorsten Joachims has done a lot of work for example. Thankfully we’re not all stuck on PageRank.

    PageRank is also flawed. It has been changed and changed and changed since that wikipedia entry and those changes will be secret of course. Google work diligently on their system and if other researchers have found flaws you can bet that they are aware of them and are fixing them.

    InfluenceRank algorithm (NEC laboratories) looks exactly at that, PageRank for social stuff. It’s an interesting paper.

    “Social relevancy ranking may become the next big thing after PageRank.”

    No, it is already in research and has been for some time now. Also it’s not really the “next big thing” and there has been a lot more research between PageRank and now, thankfully. Did you mean for SEO’s/online marketing people?

    Look at “Personalized Social & Real-Time Collaborative Search” from 2007 for example. “Empirical analysis of an evolving social network” from 2006. “Finding High-Quality Content in Social Media” from Yahoo! from 2008. There’s been a lot going on for a while.

    Right, time to turn off the iPhone and enjoy the beach 🙂

    • @cj thanks for your in-depth response. I’m by no means a search algorithm expert. I wasn’t talking about the ‘best’ algorithm in terms of delivering the best information. Google is simply the most widespread and more important economically successul in that space.
      It is likely flawed and biased, buy I do feel that social media has proven that it is easily played and cheated. The content is often very inreliable and when you take the real-time factor onto account, there is less calculation spent on determining quality, trust, reputation etc.
      Social relevancy ranking is driven by economy not by a need to serve us the right data at the right time 😉

  3. Pingback: InfoBore 17 « ubiwar . conflict in n dimensions

  4. I don’t think it will be that bad. There is still the effect of people discovering things for themselves. And people finding out that that car that was supposed to be the best, actually isn’t that great and they go off to find a better one.

    I also have to disagree on the reliability of the Google PageRank. It is not as good as it looks. It takes the most popular content to the top, instead of the best, as it has no way of determening what the best content is.
    I’ve implemented several search systems for large organisations and I’ve seen a lot better search algorithms then that.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s