As we (akrowne, jcorneli, rspuzio) discussed on the phone, it would be nice to make a "code market" for feature requests related to the AM projects. This page describes and advocates the "marketization" task.

Supporting Documents

• PlanetMath Code Bounties offering up to $600 per year per person for "odd jobs" for PlanetMath.org.
• public markets summarizes the general idea and relates to other "theory" documents on the wiki.

What needs to be done

• Better document the feature requests in terms of "what's needed" and "who says?" pairs, together with "what's offered" and "who's offering?" pairs. (For short, we'll speak of needed and offered data.)
• It would be nice to have some system for keeping track of which features have been documented in this way. For example, we could mark items that have some feedback of the above sort with a * on the Feature Requests page. However, as certain feature requests become better populated with needed and/or offered data, that will may change their do-ability and we may want to use some other tool to distinguish between feature requests with different sorts of populations.
  • For example, a feature request that has some heavily weighted "needed" data (e.g. "jcorneli says we'll need about 12 full time person-years to get the HDM to solve grad-level math problems, but he can only offer 4 person-years over the next 4 years") might show up as one type of "help wanted!" item;
  • whereas items that are heavily weighted with "offered" data (e.g. "PlanetMath is offering a $200 bounty to anyone who can satisfactorily complete this task") would also show up as a "help wanted!" item, but of a somewhat different sort.
• Hopefully we can find some suitable formalism for dealing with the difference between these kinds of feature requests, and expressing the relevant data (in summary form) to interested consumers of such data.

Just for example, I think that the current feature request has this profile:

Profile

Needs

General discussion of the problem and development of a suitable basis on which to further decompose needed/offered data (for a total of 1 person hour for each of 3 people, presumably "the usual suspects", to be put forth asynchronously over the wiki), then, on average, 15 minutes per item to create an initial profile for each, optimally followed by another 15 minutes of checking/review by some other party. --jcorneli

Offered

I'd be happy to participate in the initial discussion I mentioned above, and after that engage in a test run of profile-creation for some (say 5 to 10) of the feature requests that I have personally created. Also, I'd be happy to review a corresponding number of profiles created by someone else. --jcorneli

Feature Requests