The TSIS rating system is intended to optimize the value of the information in the system.
TSIS will make all information subject to a “value” rating, along with flags for disinformation or conflicts of interest, and so on. This applies both to original information and to critiques of that information, and any further derivative.
The TSIS rating system will optimize value in two ways. First, it will recognize (and eventually reward) great information, and the users that provide it. Second, it will guide other users to find the best information for their purposes.
The first goal of value ratings is to provide an incentive for providing great information. Gaining an audience will require good ratings, which will require genuine effort (or great talent). If you have great information with high value to a lot of users, you will stand to benefit accordingly. If you have crappy ratings you may as well not publish.
The second goal complements the first, namely to guide users to the information that they find most valuable. In particular, users ideally want to know the value rating that they themselves would apply if they read each item. This is a prediction problem.
TSIS will permit users to control how ratings are used for them, to filter and rank the information they are presented with. It’s in their interests to pick the best value prediction method(s), so it makes sense for them to make that choice.
TSIS will offer several choices, because there is no one optimal approach, and will allow mixing choices.
- One choice will be to let the system learn or tweak the user’s interests based on their actual ratings, using (something like) back propagation neural network learning.
- One choice will be to select one of several predefined ways of aggregating ratings.
- For example, TSIS should be able to select which other users or user attributes make better predictions for a user.
- For example, research shows that those best able to predict the ratings of others are much more reliable predictors of information versus disinformation.
TSIS will continuously research and improve on these smarts.
Although some people may prefer to operate in echo chambers of their own ideas, and they’ll only want to consider ratings of people who belong to their tribe/ideology/religion. We can’t help that, but many of us will be interested in algorithms that predict which information disagrees with us that we will nonetheless find the most valuable. The dream is to make it so easy to find valuable information that crosses tribal lines, that people will be tempted to explore.