Bill Guindon
5/3/2005 3:45:00 PM
On 5/3/05, Gavin Kistner <gavin@refinery.com> wrote:
> On May 2, 2005, at 2:39 PM, Hal Fulton wrote:
> > I seem to recall there was some discussion here of
> > people paying small amounts for small pieces of code
> > (a la rentacoder).
> [...]
> > Does this concept seem interesting to anyone? Worth
> > discussing?
>
> I've been wanting to do something like this on an IRC front for a
> long time, but the issues of Trust have made it hard to figure out
> the details.
>
> It's an exchange of goods where neither person trusts the other, both
> holding on to what they have while grabbing for the other person's
> goods. When it's something physical, you don't let go of what you
> have until you're sure that you have a firm grasp on what the other
> person is offering.
>
> But if it's information, I see two choices:
>
> 1) You let the person offering the bounty review the answer and
> discover if it's valid. How then do you prevent that someone from
> getting their answer and then saying "No no no, that wasn't what I
> wanted at all. I'm keeping my money (and the information that's now
> in my head)." ?
>
> 2) You force the person offering the bounty to pay up before seeing
> the solution. What then do you do if the solution is "Ha ha, you
> suck, I've got your money now!" ? (That case is easy to resolve by a
> third party, but what if the solution is real code ... how much work
> do you want to do diving into each solution and determining if it's a
> 'perfect' match?)
>
> Hrm...what if the answer is precise specifications, in the form of
> unit tests? What if the person offering the bounty is responsible for
> providing a clear set of specifications AND unit tests for the
> interface, and the System runs the unit tests against the solution to
> automatically verify that it's valid.
>
> With the IRC model I have been thinking about, you'd need to
> bootstrap the system and get to a point where everyone involved had a
> nice history of Trust rankings, based on grades of their solutions,
> number of solutions, number of disputes, and so on. (You' d also have
> those rankings distributed over a wide variety of information
> topics.) But that's a general description of the end result, and
> glosses over the details of how to get there (and assumes that such a
> system makes 95% of the people using it happy).
Well, as was pointed out recently, the 'Peer Review' feature on
RubyForge is pretty much just sitting there at the moment. Seems a
logical place for eBay-like feedback. Of course, it's probably also
ripe for abuse. Maybe it could be tightened up somehow.
> --
> "When I am working on a problem I never think about beauty. I only
> think about how to solve the problem. But when I have finished, if
> the solution is not beautiful, I know it is wrong."
> - R. Buckminster Fuller
>
>
--
Bill Guindon (aka aGorilla)