Re: barriers to deployment of web annotation?

Ken,

I agree that abuse, self-interest, etc, are big potential problems for any large-scale annotation system. I also think that human review at some level is the only solution.

You describe this solution as having "huge costs and scalability issues". Am I right in assuming that you're imagining a regular staff (paid or voulnteer) who comb through the posts? What about a more distributed form of human review, where users rate posts for relevance, etc? This addresses some of the issues of cost and scalabliity (and, of course, introduces some of its own problems).

I wrote a rant about this a couple of years back. It's at:

	http://www.web.net/~misha/annot.html

The part that specifically addresses collaborative filtering of comments is at:

	http://www.web.net/~misha/annot.html#filters

Hope these provide some ideas to chew on.

			- Misha





At 02:17 PM 12/8/98 -0800, you wrote:
>dlaliberte@gte.com wrote:
>
>> Jakob Hummes writes:
>>  > Rolf H. Nelson wrote:
>>  >
>>  > > 1.   Lack of user agent business model:  [...]
>>  >
>>  > Maybe, but then: If third party annotations exists, why should they
>>  > hinder them?
>>
>> I have heard, from someone who was in a position to know, that Netscape
>> did not think there was enough of a market to justify supporting
>> annotations.  Furthermore, even if there were third party annotation
>> services, the big info providers whose pages would be annotated without
>> their control would not necessarily like this situation, and they are
>> the ones who ultimate pay a large portion of the bills (well consumers
>> really do, but they don't have much control).
>
>We have found that annotations, such as our new What's Related feature, can be
>very tricky. Large content providers are unhappy with the prospect of others
>making money off of their content (do you remember some of the suits associated
>with "framing?"). The idea of a simple tool to take users away from their site
>doesn't excite them either.
>
>There's also the problem of creating annotation content. An open system for
>users to submit annotations will quickly create something as messy as the
>current domain system -- e.g., an up-for-grabs space that will get exploited by
>people trying to steal traffic. You just can't trust people to stick to rules
>about keeping annotations relevant. Note that this can create legal exposure
>("why are these porn sites related to my site?") for the company providing the
>annotation infrastructure, and leads to a system that has low value. The only
>way to prevent this is to have humans review each entry, which has huge cost
>and scalability issues.
>
>We think that annotations, such as related sites, news stories, reviews,
>company profiles, etc. would be very interesting to consumers, however, we are
>still wrestling with ways to build and maintain the databases. Opening up the
>databases is tougher yet. We think there's a possibility that the new
>volunteer-based Netscape Open Directory (formerly NewHoo) may solve these
>problems (scalable, open, checks and balances).
>
>--KenH
>
>
>

Received on Tuesday, 8 December 1998 18:08:09 UTC