Giulio Cesare Solaroli e-mailed me to say he enjoyed my blogpost on his and Marco Barulli’s computational reputation model for blog comments, but to point out that I’d misconstrued several details regarding the interactions among functional components in their model. I refer you all to http://www.clipperz.net/ (Giulio and Marco’s blog) for further details on their model (as those details get posted).
What I’m doing in this blogpost is responding to the core assumption of Giulio/Marco’s computational reputation model, as stated by Giulio in our e-mail thread:
- "What we hope, is that the merit of the comment and the reputation of the author could be some how (indirectly) bounded. Smart people tend to be smart; troll tend to be troll. This regardless of the context where you observe them."
As I responded to Giulio in the e-mail thread, here's my philosophical perspective on this issue:
- A person's perceived "intelligence" depends totally on context. Intelligence is primarily the capacity of an individual to respond (continually, appropriately, effectively, articulately, and successfully) to various challenges (tests, tasks, and problems) that the world (fate, society, colleagues, teachers, friends, and adversaries) place in front of them. The evaluation of the success of a person's ongoing/evolving "responsiveness" to the never-ending parade of new challenges is that person's "intelligence." The persons doing that evaluation include the individual him/herself, plus the person's family/friends/colleagues/contemporaries, plus the person's posterity (descendants/historians/etc.). The context for the "smart vs. stupid" determination, then, is the entire frame of reference that involves diverse challenges and different evaluators. Nobody is inherently smart or stupid--they must continually "prove" themselves as one or the other, and could just as easily (on their next challenge, in the eyes of their next self- or external-evaluation) flip-flop toward either pole. The bottom line is that people say smart things on some occasions on some topics, and stupid things on other occasions/topics. People tell me I'm smart, but I'm quite aware when I've said or done something stupid (or my wife makes me aware of it).
- The evaluator of a person's intelligence may be particular individuals, or a community of individuals, or a particular individual deferring to the collective/received opinions of the community (contemporaneous/posterity). Your model is based on the latter intelligence (reputation)-evaluation model: a particular individual (blog author) deferring to the collective/received opinions of contemporaries (other blog authors, as filtered through a "reputation manager").
As I stated in my recent blogpost on their model, I base my decision to reference somebody else's inputs in my blog on whether they pass a certain "intelligence" challenge in the eyes of one particular evaluator: the "it's interesting to Jim" test. I'm not so interested in whether they pass the "intelligence" challenges of other evaluators (such as a circle/community of blog authors). It's not that I necessarily disparage the opinions of other blog authors. But they run their own fiefdoms, and I run mine. I'm president, king, emperor, and grand vizier of my own idio-domain.
I'm always skeptical of "received community opinion" (aka "reputation"). As I said in my blog this past November:
- "Reputation feels anti-governance, hence unfair. It feels oppressive. It’s the collective mass of received opinion, good and ill, weighing down on a particular identity. It feels like a court where the judge, jury, prosecuting attorney, jailer, and lord high executioner are phantoms, never showing their faces, but making their collective force felt at every turn. It feels like outer appearances, not inner character, ruling our lives."
Giulio responded with another e-mail in which he expressed some doubts about whether “reputation” is really the concept they’re trying to capture in their model. I said that, in the context of their application, I don't think "reputation" is the right term or concept for the assurance level that the "reputation manager" (wrong term for that functional component) is asserting with respect to a blog commenter. What the blog commenter asserts (across one or more comments made to one or more blog authors) is their "commentary" (the sum total of their comments, as an outward manifestation of their analytical and expressive powers).
I proposed that Giulio/Marco think in terms of the following roles:
- Blog authors: These are as Giulio/Marco define them. But it's important to recognize that a blog is simply a stream of commentary from one or more blog authors (e.g., Marco, Giulio, et al.) and (optionally) blog visitors (who may or may not be able to post their commentary to a blog author's site, or, if invited, must be approved by the blog authors prior to posting).
- Blog visitors: These are Giulio/Marco’s "blog commenters." They don't own the site that they're visiting. They're just guests. They knock on the door and may or may not be invited to post directly.
- Blog authors: Yes, this is a second role (reviewing submissions from blog visitors) that blog authors perform. Blog authors may also publish their "reviews" (or thumbs-up/thumbs-down decisions) to a blog review hub.
- Blog review hub: This is Giulio/Marco’s "reputation manager." It's job is to compile, aggregate, weight, and score blog reviews from various blog authors pertaining to various blog visitors.
Blog review hubs don't score blog visitors' "reputation." They score the community (of blog authors) reported evaluations of the quality of blog visitors' submitted comments.
It's not reputation. It's "commentary quality" (of blog visitors) that's being scored by a blog review hub.
And that's what blog visitors (that subset of them who simply visit to read commentary) are implicitly scoring through their browsing/attention and return visits. Do they respond to the quality of Marco/Giulio, Jim Kobielus, Phil Windley, or other people's blog-asserted commentary? And to the quality of the blog visitors who those blog authors have opted to let post to their blogs?
The blog review hub must always be scoped to a particular commentary community. Any measure of “commentary quality” is always relative to the yardstick (i.e., set of values) that a particular community (e.g., technogeeks, feminists, right-wingers, conservative Muslims) holds in common. It’s not enough to scope it to the “blogosphere” (as if that were a community). Your blog review hub may be associated with a particular community of interest. Or a particular congregation of idiots. As blog author, you choose which blog review hub(s) you wish to federate/affiliate with. Whatever clique you click with.
Essentially, then, a blog author may wish to consider a blog commenter’s commentary quality score, as reported by a particular blog review hub, prior to posting that commenter’s comment to their blog.
Or take the simpler, more direct route. Actually read the submitted comment. Then hit “post” if so moved. Or process the comment through your own gray matter and blog on it.