(TinyURL.Com/RevTre = "reverse tree")
.
The basic idea is the inverse of a "phone tree". Instead of one person generating content and relaying it through an ever-expanding tree of volunteers, lots of volunteers collect information and filter only the very best up to the next level, so the one person at the top gets the best of the best of contributions and doesn't have to waste time dealing with the majority of fan mail etc.
In addition to merely filtering submitted content, content can be improved as it rises from submission-leaves toward final-root, with feedback backward not just to say why submission rejected but to show how it was improved to make it more acceptable to pass further. In fact this mechanism can be used in education where less-expert students submit material and more-expert students repair it and tell less-expert student exactly what repair was performed, and thus less-expert students get coaching from more-expert students.
Examples of uses for this mechanism:
- Errata for Web pages (URL and context of erratum) so that WebMaster needs to look only at verified errata reports;
- Contact with famous/busy person, such as popular actress or political candidate, or CEO or ombudsman of major corporation, or person who live-tweeted major news event;
- Witnesses to news events, such as shooting in school (large), thereby in effect creating a "newswire service" that consolidates best of the reports;
- Filtering first-contact e-mail to people who hate spam, such as myself (large), or filtering ('moderating') public comments to a 'blog' to keep out profanities and spam;
- Filtering a "group feed" into something like Twitter, where only the best gets posted to Twitter, spam and worthless chatter drop out early;
- Collecting reports of spammers on newsgroups or Twitter and filtering these reports to produce list of worst spammers that we'd most like to get rid of, a better-organized group effort similar to what FrankleeMiDeer is trying to do all alone;
- Filtering the many tweets of the people you are following or would like to follow on Twitter, so that you spend your time reading only the most interesting tweets;
- Filtering suggestions about what news stories are really important but not yet appreciated by most people, which therefore should be given greater attention by investigative reporters and news editors and the general public. For example, it shouldn't have taken 10 years for reports of birth defects resulting from thalidomide taken by the mothers to get recogized by the media;
- Filtering submissions to @BobMankoff ! (the carton editor of New Yorker magazine), thus taking most of the load off his assistant and himself;
- Distributing load from single person, such as Robert Langer, answering questions from millions of people: If somebody early in the tree can answer the question already, the answer goes back immediately, without wasting the time of people closer to the root. Only questions almost nobody else can answer get all the way to the root person to be answered there. Aided by a FAQ database and query engine, as soon as one person has answered a question from personal knowledge or research, then fed the answer towards the root for approval, and it's then added to database, later it's easy for anyone knowing appropriate keywords to look up the answer to similar questions, so questions for which the FAQ is sufficient get answered very early in the RevTre;
- Learning a text-based skill such as translating from one language to another: Computer makes stab at translation, then translation-pair is fed to beginner student at leaf to check if it looks correct and if not then repair it; Successively more-expert students closer to the root vet the work of more-beginner students, make corrections, show more-beginner hir mistakes; at the root the translation pair is almost sure to be totally correct, so it can go into a translation-pair database used both for student research and training material for the automatic translation software.
The trick to get people to work hard at pleasing the end-recipient / target (person or newsfeed etc.) by their efforts filtering stuff for that end-recipient to see, is to use a truth-futures market whereby people put a "price" on each item being filtered, that "price" being the probability that the end-recipient will like to read the item. The market "price" for an item, rather than any one individual's opinion, is used for placing the highest-value items in the target's BestInBox. The funds that can be earned for good estimates, or can be lost for bad estimates, are the incentive for good filtering on behalf of the target (person or newsfeed etc.).
IMO a reverse tree with a reward system for good work is the best way to accomplish a crowdsource-curated feed.
.
UP
.
.
.
.
.
.
.
.
.
.
These external links are to regular Web pages that are designed for full-screen Web browsers, not appropriate for viewing on cellphones:
tinyurl.com/f24sho
indirects to
Germany school shooting: thanks to Twitter, France 24 gets live account after 52 mins | The Observers
First detailed draft of reverse tree with myself as the target.
.
.
.
.
.
.
.
.
.
.