Belief calculation : old

In Believing evidence we explain what the possible beliefs in a evidence-fact, and its relation to the mainfact are, and what it means.

Source links (i.e. (S -> F) or (S !> F)) are handled in an analogous way. Yes, this means we can allow users to have (believe, disbelieve, or be uncertain) opinions about sources now. I would still like to reserve the right to allow users to do hard rejects for sources in their personal views but that’s something we should kick down the line since it has major product implications and may be unnecessary.

Authority of sources is simply the number of Facts that use them in SourceFacts. User opinions about sources shouldn’t add to the authority of the source since we want a purely structural value for source authority.# And, higher authority sources can be expected to be used more in practice. For example, the first time a source is used it has authority 1.0 and contributes that to the SourcedFact (new word) that references it.

Users still sit outside the system a bit for now since at this point we haven’t figure out how to allow users to discount other users, endorse them, etc. This also prevents a lot of gaming in the system and could even be more democratic. As with sources, we may want to support hard rejects for personal views but it’s unclear whether that’s necessary for any version.

What about (F1 == F2) equivalent facts? These are tricky and we could get some very nasty recursive unbounded calculations so for now I recommend we don’t use them in the opinion calculations though we can record them as actions if we want. In other words, we don’t propagate credibility across them yet (later we will) but we can surface them to users and allow them to choose which one they work with, believe/disbelieve the link, see the links, etc.

In general, this is a good way to handle hard calculations early in the product life -- simply work them at the UI level so they are visible and manageable.