A VC« MP3 of the Week | Main | The Lone Star Cafe »
Reputation Services
The Internet is a fantastic environment for delivering communication, commerce, and content services. It’s also an incredibly efficient way for users to access these services.
But its also a place where a lot of bad stuff exits. My Internet axis of evil list has viruses, spam, spyware, and phishing on it. I can confidently predict that there will be other maladies joining that list.
So it’s the best and the worst. And as a result of that, there is a new class of web services being built called reputation services. Some of these have existed for years. eBay’s reputation ratings were a key component of the success of that platform. Amazon’s rating system for reviewers is another excellent example. But in many ways, these are web 1.0 implementations of the idea of a reputation service. I think we’ll see a lot of innovation and development around this concept over the next five years.
Here is my list of things a reputation service needs to do:
1) Identify things you can trust. It’s not just about people, but people are an important subset.
2) Identify things you can’t trust.
3) Incorporate user feedback.
4) Incorporate other data that is relevant.
5) Allow users to set their own standards. For example my idea of a bad review may be very different than yours.
6) Provide for syndication. I might want to take your reputation and make it my own and add more user feedback into it.
7) Provide for extensibility. I might want to take your reputation service and add my own layer of reputation service on top of it.
I’d be interested in hearing what you all think of this list and please let me know what I missed.
November 8, 2004 Venture Capital and Technology | Comments (4) | TrackBack (0)
TrackBack
TrackBack URL for this entry:
http://www.typepad.com/t/trackback/5934/1386838
Listed below are links to weblogs that reference Reputation Services:
Comments
Fred,
You may want to check this company http://www.canalstreettalent.com/ that is doing similar stuff.
Posted by: Jawahar Mundlapati | Nov 8, 2004 11:48:06 PM
Like all interesting things, it's important that the initial reputation systems do one thing well before trying to extend out and meet all reputation needs. Note that both eBay and Amazon's services are addressing a small part of the puzzle, hence their value now. Web services may also help address this issue, where the need will not be so much to have a reputation system as much as a repository for all of one's reputation items (ie. Amazon reviews rep, eBay merchant rep, etc...).
Posted by: P-Air | Nov 9, 2004 7:39:26 PM
Hi Fred,
I like your list, and think number 1 is good: "Identify people you can trust." I'd say that having standards for people I can trust is much more reliable than trying to maintain a set of standards that describe people I cannot trust (your number 2.)
People I cannot trust keep changing, moving, hiding. The list of ways they do things I don't like grows infinitely. To me this seems analogous to letting a stream of strangers into my house and then trying to watch them all while taking notes when they do things I don't like. Then using those notes to judge the new people entering.
On the other hand in email - companies, domains and servers that don't move around and don't hide - present that as a clear measurable characteristic in itself. And when spammers emulate not moving and not hiding, they get on a blacklist or get sued - which again are clearly measurable characteristics.
I think user feedback has a weaker correlation to which people or email messages I can trust - as you noted, everyone's opinion is different. I might not trust "people over 30" or you might not trust "people who wear suits and have pony tails" or Joe over there might not trust "people with detached earlobes."
We have customers who literally do want to buy Viagra online or try to get a free iPod and they move that mail out of bulk into their inbox. Conversely, people take jokes sent by friends and even RMA numbers in email for $500 products they need repaired and mark them as Junk or Spam and never open them. And I believe we've seen a recent incident of far left and far right organized groups gaming the other's email reputation via organized junk mail reporting.
Automation of user feedback as the input to reputation systems leads, in my opinion, to oversteer. I suggest that sticking to who is stable, has been around a long time OR is clearly identifiable - and is not on a well-run blacklist or charged with breaking privacy / spam laws - is a reputation method with less oversteer.
Why do I think blacklists are ok but not user feedback? The best blacklist operators are highly skilled and careful people sticking to a set of clear standards and with many years of experience doing forensic analysis on email messages.
Thanks for your article, I enjoyed seeing a mention of reputation systems after running one for years during which time there wasn't even a category or name for it.
Posted by: April Lorenzen | Jul 31, 2005 7:57:21 PM
Hi Fred
What I liked about point 1 was the reference to things and not just people.
I can see HUGE interest and activity in online reputations, but hardly anything on trusting websites. Maybe its not really required, and people will always search around even with such a system in place, but it could be something to think of.
cheers
Gaurav
Posted by: Gaurav Rekhi | Jan 8, 2006 8:00:12 PM
Main RSS Feed of All Posts
VC and Technology Posts Only
