I've been following Peer Index for a month or so. Peer Index is attempting to "help you discover the authorities and opinion formers on a given topic."
The idea is to use the world of social media (blogs, twitter, facebook, linkedin, etc) to determine who are the "authorities and opinion formers" on the web.
I like the idea. The execution is the hard part. When I first tried it out about a month ago, I thought the service was too thin. Too few topics and too few "authorities" in each topic.
Sometime in the past week or so, they added a bunch of new topics and a lot more authorities in each topic. And they've added a bunch of new ways to navigate the service. I like where they are headed with this.
In a perfect world, you could enter literally any topic, like "sushi in london" and find out who the experts are. Then you could follow them, read their opinions on the topic, and possibly even contact them.
In order to reach that perfect world, I think they are going to need to open this up. Let the "crowd" source the topics and the authorities and then use their algorithms and systems to maintain the topics and rankings. At least that is my gut on how this would work best.
In any case, this one is worth watching.
And they’re a happy Superfeedr user 😀
uFollow is creating an index to the blogosphere on the author level. Have a look at http://www.ufollow.com.
This is a challenge Fred. I have checked out influence sites for Twitter. They are horrible, and often I think they are hype/scam machines, even Edeleman Digitals new one, context is always missing (Search suffers from the same handicap). I will have to check out Peer Index. Immediately I can think of great uses for such an offering, especially in the realms of Science, R&D, anything technical etcI really like your idea for Travel/Restaurants etc. But we have forms of this already with so many websites and networks. And we are individuals. So I think each expert really needs to align with your tastes. I might want to know a great Hard Rock Club in London, but if someone’s favorite Band is Bon Jovi I wouldn’t want to take their advice. (In their mind Bon Jovi is hard rock, but not mine). I used this example not to offend anyone into Bon Jovi but because it was an easy analogy! Context is key.If there is a way to filter properly this could be a winner.
It looks like someone who shares tons of mediocre content that is already a meme gets ranked higher then someone who infrequently originates new content.Loud != authoritative.Frankly I think Disqus does what these guys are trying to do better than peer index does.
That happens when you try to rank automatically because quality is difficult to read for an algorithm and it’s not an objective thing. On the other hand, if you try to it manually it’s not gonna scale, although for some niches it can work great (like here).I think that trying to scale the niche model is gonna be more successful in the short term, but it should get some automatic tuning to make things really comparable (for example, in some blogs likes are more usual than in others).
yeah, anyway the answer is 42
Sadly, being loud does resonate with having some authority. :(Some people do resonate loudly because they’re being listened to though – but not because of their volume.
what about disqus plus peer index?
An interesting idea, I’d love to see them work on that.
Is a “leader board” the best way to interpret authority? One obvious downside to this is: Those who respond to gaming incentives like leader boards are the same people who know how to game a leaderboard. As a longtime watcher of attempts to measure authority, I’ve noticed dual dilemmas: Who is and authority for me, may not be an authority for another (for example, someone’s music opinions and tastes for a major music genre may make that person appealing to a large audience, but not to my niche tastes).Secondarily, all algorithms have to start somewhere and those that start with Twitter (and, especially having to do with early-adopter topics like tech or music) are rewarding inertia that may not have anything to do with “authority” but with being a part of a “crowd” that shows up first on any new shiny thing. What is the point of authority, anyway?This morning, I was trying to remember the name of a travel scanner someone mentioned months ago on their blog, Twitter or who knows where. I found the name of what I was looking for via a Google search and the “social search” feature they’ve added. But then, using GetGlue (which I know is in your portfolio, but this is not a “shout-out” for that reason), I was able to see who — among people I know — had either visited, ranked or reviewed that specific product. In other words, I could find “authority” among my little crowd, as it relates to a specific product, not “scanners” or “electronics” or “technology.” I think “authority” is most helpful to me (as a user) when it is influenced by my “connections” (those I trust or who share work or personal interests with me) and is provided in the context of search, decision-making or finding a solution.
Hi Rex,You’re absolutely right about the limitations of any given data set from which to compile authority. Tenure boards look at peer reviewed journals (and often don’t consider an academic’s brilliant blog posts) when considering an academic applicant. And applicants who don’t publish in the right places, however brilliant there ideas, will tend to get overlooked. Clearly, we don’t want to limit ourself to our data that is visible through the Twitter flow–if there are useful predictive signals elsewhere–it is one reason we ask people to (optionally) provide LinkedIn connections.You’re also right to identify that there are multiple use cases for ‘authority’ some of which end up being distinctly personal. From our standpoint, the trend — that there appears to be both a fragmentation of authority and a better ability to track ‘lower’ volume signals of knowledge — is inescapable. And our task is to fing the right mix of services to a world when authority isn’t simply defined as having a column in the New York Times, covering a sector for a sell-side bank or being the closest friend to hand who owns a dSLR.As always, feel free to reach out to me aa [at] pi.mu with observations, criticisms or ideas.
Thanks. And despite being in the context of this post and thread, my observations were not intended as a criticism of your approach, rather with an observation of the broader issues of measuring “authority.”My point, in general, is that far more factors (and data points) will need to be considered in scoring “authority” if it is to be helpful in the context where most users will find it of greatest value: in the context of decision-making. If the point is merely yet another topical listing of “authority” and mixing in some gaming-inspired incentives to the user-experience, the algorithms may be different, but at the end of the day, it’s just an enhanced version of things we’ve seen dating back 15 years and, in the past decade, everything from Technorati to Digg’s http://wefollow.comAgain, I don’t mean for this to be a rant. It’s not. I’m happy people like you are trying these different approaches.
yup. i wrote about the taste neighbor issue a few years back. it is one of my favorite posts.http://www.avc.com/a_vc/200…
That’s a great post … and it made me so hungry for a fish taco, my wife and I heading out for some now. ; )
Nobody has solved the authority game in a “fractal scale” what I mean is that is easy to recognize authority at a big scale (techcrunch, nytimes, avc) but complex in a small group. Even Google with it’s pagerank fall short with niche authorities.One of the reasons is that authority is not 100% correlated with links or following/followers, if you don’t want to promote yourself you loose/not-gain authority in this market, also is Twitter/facebook who have more information to work on this. In the last 2 years I found more difficult to search niche stuff on Internet while in the past Google results where enough.I think something really complex is happening.
There are ways, and they’re simple, but the info available isn’t there, yet.
Understanding topic based authority is one of the most exciting challenges on the social web. You have companies like Peer Index, Peer Pong, Topsy and us here at Klout all attacking this problem from different angles. The convergence of technology and consumer behavior where people are comfortable posting their opinions and sharing their expertise with the world and the data is available through api’s and other services is allowing this space to take shape very quickly.We first started published topic based social authority rankings almost a year ago. Looking back at the quality of those results it’s embarrassing versus what we are seeing now. As you stated, this is a huge challenge but I am looking forward to seeing how all of us in the space catch up with the business opportunity at hand.
i like Klout too Joe
Joe, thanks for the shoutout on PeerPong 🙂 Definitely agree that’s theres a number of ways to approach the problem. Using NLP/semantic analysis on social stream data including Twitter, FB, Linkedin and other data sources, our index now covers 4M people, growing at 1M a month. We actually already can identify authorities for “Sushi in London” and for any combination of key terms.
Pretty neat. One of the TechStars Boulder 2010 companies is doing a similar thing (Spotinfluence – http://www.spotinfluence.com). I asked them to get you a private beta login to play around.My first complaint about Peer Index is that it looks like they are batching their data instead of using real time. Specifically, the number of twitter followers I have is a few weeks old and low by a couple of thousand. This will miss real-time / trending / high velocity influencers.It’d also be great if you could compare influencers – e.g. search on Brad Feld and Fred Wilson simultaneously.
We are deliberately batching calculations. Our take is that authority isn’t something that changes on a minute by minute basis, and if you’re time horizon is too short, you will end up calculating the wrong thing. The reason you or fred and up high in our vc rankings is that you sustain high quality, desirable behaviours over time, not because you were the first with a particular funding announcement.We did indeed run our authority calculations very frequently– and decided it didn’t give us anything extra. Your profile should be updated when you claim it. If numbers were off kilter, it could well be the heavy memcache layer we run.Figuring out trending influencers in a purely realtime context is a possible extension but isn’t out focus right now. There is a beta works company called socialflow (afaicr) which does offer this – focusing on helping marketers get the most buzz out of their messages.As for comparisons–this is one of the things we offer and has to be done *a lot* better–but you can do some comparisons: http://pi.mu/c/1r
hi fred- given that your writing about peer index, i take you have, or are about to invest in this company? if no, then congrats on writing about a company thats not in your portfolio.
i have no current plans to invest in Peer Index. i also have no current plans to invest in a family CRM service which i wrote about yesterdayi just went back and looked at all the posts i wrote this month. fourteen so far (one per day)seven were about companies or services. four were portfolio companies. three were not.the other seven were about other things.
thanks for the facts and i stand corrected! glad to see your more broad & unbiased than i originally thought. maybe the times i check in just happen to be when your chatting about your companies.
That can be gamed pretty easily with some time and understanding and effort, and accuracy is difficult to determine. But if someone’s capable of doing a good job understanding it, they likely can put that effort somewhere more productive – until the Peer Index or others become large sources of traffic and influence where it becomes very worthwhile to game it… mind you, there are some things that can be put in place to deal with this, but you’re right, part of it would include being open.Oy, they need to buy the .com
Authority comes with a dichotomy: outsider versus insider. And outsider can definitely be an authority on a subject, but he does so in a way much differently than an insider would. And by virtue of the fact that an outsider is an outsider, it makes the issue even harder. I’ll try and explain.Insider authorities are easy to identify –but not necessarily easy to access. Steve Jobs is a great example: if you want know something about Apple, he’s the authority. But (a) he’s not likely to tell you and (b) even if he did, it comes with insider spin as baggage. But yeah, if you want an authoritative answer about something Apple, his mind would be the one to read. Easy to identify.But outsiders? Very hard to identify.First, they need to develop reputations, which requires that they’re outwardly discussing the topic about which they’re authoritative. John Gruber, clearly, fits this. But what about the hundreds of others who obsess over the same?Second, they have, relative to an insider, very little information. So even while their conclusions are correct based on the information they have at hand, they’re incorrect given the full picture. See http://kottke.org/10/07/the…And finally, there’s enough of a gap in this information that two perfectly rational outsiders can come to starkly different conclusions. So instead of having two authorities on the subject, you end up with none.
Gruber is more passionate than authoritative.Opining loudly while being wrong as often as he is does not seem very authoritative to me.
Fred,Thanks for this! It was a great weekend surprise.We’re thinking hard about the questions of granularity — the main issue being less of one to do with our platform and the technology–but more to do with the domain expertise required to provide some kind of meaningful scoring around a hyper-niche topic, where we might not have any expertise.We also want to learn from companies that have been super-successful through transparency and authenticity (like Zappos or StackExhcange) so we’ll take steps in the next few days to both expose the dozens of topics we have built (but not publicised) and to invite discussion and debate on the topics we ought to create. It won’t be the ‘last solution’ but certainly a substantive way to get rolling.Thanks for the feedback.
being transparent and authentic is so important azeem
Def. interesting…I think there are a lot of people poking around at this realm but having difficulty nailing it in any way that will properly scale (and improve with scale rather than fall apart with scale).http://socialflow.com , while it probably doesn’t map directly in people’s minds because it’s approaching the problem from a different direction (but I feel like fundamentally they are attacking the same problem)…I do like what what they are trying to do as well (maybe a combination approach that meets socialflow and peer index in the middle somewhere?)
Nice article! I like that so many people have opinions about the influence space. I’m one of the co-founders of Spot Influence, which Brad mentioned (your invite has been sent).We believe that contextual or niche influence is based on the Reach, Relevance, and Impact of individuals within a community. Reach being how far a persons thoughts and ideas can propagate within the community. Relevance is how frequently a person shares thoughts and ideas relevant to the community. Impact is when a person shares their thoughts and ideas, does anyone in the community take measurable action. Combined, they define how influential a person is within that community.We’ve been working on it for about 9 months and are prepping for beta, it’s been a really fun ride. Like Joe and Azeem we learn more about this problem every day and how we can do it better. Thanks for this post and sparking a great discussion!
When painting with a broad brush as such, it’s tough to provide precise determinations of influence or authority. One of the issues is that Twitter/the Socialsphere reward effort whether authoritative or not. This is an area we (my company) have been thinking about as well, at the topical level and we think it would be more useful to have a tool where the user can manipulate what they deem as the factors that matter to them, rather than having a pre-sifted ranking that’s too broad. And knowing how these formulas are calculated is important. The socialsphere is relying on fairly simplistic parameters like # of followers, # of re-tweets or Incoming Likes, etc…all quantitative and gameable metrics. Where is the Qualitative aspect? I’d be curious to know how it enters the algorithms.Following the “Sushi in London” example, I went to Peer Index and wanted to know who is an authority on Semantic Web in Toronto (I should be there) or “Cloud computing security enterprise” ,- nothing. We’d get better results searching Google or even searching Twitter for that matter.I think we’re still in the experimentation stages for these types of social metrics. They make for interesting vanity checks, although it would be useful if the Long Tail of authority reporting could be made more visible somehow and with accuracy. To a non-expert (i.e. consumers at large), anyone that knows more than they do is an expert, but among experts, expertise is harder to rank.
[This was meant to be a Reply to Dave Angulo]Thanks for sharing part of your algorithm (which is good to know, of course). I’m curious by how you define “anyone in the community take measurable action”. Are the community’s actions similar in some ways to the originating actions. And as I hinted to in other comment below, is there a qualitative aspect that come into play so that effort doesn’t override the entire score. (btw- I requested an alpha in case you have the “authority” to bump me in 🙂 )
morningstar for people.
we need people to replace morningstar
what i meant. i suck at disqus pitches although azeem understood way back when I mentioned it to him ion an early discussion
Hey, I’ve got a top 50% badge for entrepreneurship.From all those AVC tweets for sure.We won’t talk about my authority score. 🙁
There is an irony to this. Back in the early Internet days, search was Yahoo (or AltaVista or similar), and it was largely manual: the index was generated by a bunch of “insiders” on where the data was, a model taken to the obvious next level by About.com. Then came Google with its automated PageRank (Backrub?) and it became impossible for manual indexing / search to keep up (classic case of a process ripe for automation).Now we are talking about a service that uses automation in “crowdsourcing” to point to experts, sort of Back to the Future. There is some irony here. Using engineering analogy, this feels like the return to server-based processing that Web apps represent after years of client-server. The world really does go full circle. doesn’t it?
Hi Fred,A bit off topic, but I thought you may want to know that your email update feed to me on this post included advertising for “foreign brides” in the guise of contact ads, like – “Kitty, 20 yrs, Russia”. I don’t know how it is on the US, but in Europe these are very often run by hard-core Eastern European mafia using them to sell women into marriage against their will. I don’t know who you use to handle your in-email ads, but I’m sure this is not what you had in mind…if you’d like a screen print of the email with ads send me an email. Best, Chris
Human trafficking from EEuro — very big problem.
ughthat is google, not mei run their ads in my feedif you can send it to me, i will forward to themi doubt they will care too much though
I just requested women in business as a topic.Ultimately I’d like to not need it but until then it would be very useful to have a dynamic look into the influencers. Useful to find each other. It’s a fragmented group.
The idea of a peer index is simply amazing, the problem is the execution using technology. Peer Index seems work rather crudely, when i typed in innovation, it simply listed twitter accounts with ‘innovation’ in their username.umm…needs some refining.
There’s already a pretty good peer index: social search.
Fred you should check out pinterest.com
QDOS.com also tries to measure one’s online impact. Somewhat similar to Peer Index.
Peer Index sounds similar to hunch.com. Hunch leans towards crowd-sourcing to identify answers to a question like “sushi in London” and helps you identify authorities on a topic within the site. It doesn’t have that farther reach of crawling the web though as far as I know.http://hunch.com/london-sus…
How is this at all different from BuzzLogic?