More actions
From Ratings System
Writ large we are creating a self-improvement tool with profound social ramifications. We are giving people the ability to use the internet's vast resources and each other in a constructive way, free of the negative influences that are normally associated with online information. It would seem that a software system of such ambitious magnitude would have a wide variety of applications. So far, however, we have only explored the system at a high level in terms of factual/probabilistic information retrieval and debate/logic. Superimposed on both is a trust network where users can rate other users and the information they are receiving in a personalized way. We have further noted that media/social media is a clear application for this system. Other applications might be:
Academia: Reliable information that can come from anywhere will free up educational programs to tap these resources and not have to rely on repetitive (and expensive) in-house content creation. Education can finally achieve the decentralization that we have always believed would come but has only made the slowest of progress. One problem we will need to solve is credentialing, that is, how do you obtain a "degree" from information like this? One answer, admittedly vague, is to have online communities function as "schools" that agree to ensure that students have completed some level of mastery over the subject matter.
Educational Material: Have you ever read a Wikipedia post that assumed too much knowledge on your part, ie that was pitched to the expert? Wouldn't it be cool to be able to ask a question and get a response that anticipates your level of knowledge of a subject and pitches its content accordingly? Our system could do that because entries could be rated based on level of expertise required (simple to advanced). Since a demand will probably exist for simpler explanations, contributors could identify and meet them. Users could then adjust parameters having to do with their level and obtain a customized explanation. The system could obviously store articles based on rated expertise and call them up for future requests.
An extension on this idea would be to have questions linked off the main article that readers would create, rate, and upvote. The highest ranked questions could be answered in their own link or alert the author that this is an area of confusion that needs to be fixed.
Government: Information sites run by local, state, and national officials might be better served by allowing people with first hand experience to provide this information, either alone or as a supplement to the official site. The government site can offer a tailored version of reliable views held by those it considers trustworthy. Needless to say, such a tool could also help policymakers design laws and regulations which have been properly vetted and are politically realistic.
Social Services: A great many services exist for the poor and disadvantaged which are underutilized because the paperwork needed to sign up and maintain them is onerous. Furthermore the government agencies responsible for them are not good at providing the needed customer service. The consumers, for their part, tend not to be savvy about navigating the system because they may be old, handicapped, or otherwise incapacitated. Private companies have little incentive to rectify the problem since the profit is not that great. Clearly this is an area which would benefit from a reliable and easy to use information system. Offering preset algorithms and mathematical weights will be important for the consumers of this service since it is unlikely they will be able to manipulate these on their own.
Workplace / Business: The system can be deployed as an intranet for the purpose of information dissemination and project collaboration. It could serve as a way to mitigate the impact of cognitive biases that especially affect large workplaces (eg groupthink, authority bias). Large projects are often undertaken with insufficient scrutiny and without a proper ROI analysis. Even engineering designs are often committed to without a proper high level analysis of alternatives (AoA). Other workplace applications might include a more objective rating system for employees.
Medicine: A large problem in medicine is correct diagnosis, for which a wide network of experts would be ideal. Indeed, in hospitals today doctors rely on the independent assessment of other doctors, NPs, and PAs to come to conclusions about the patients in their care. The usual MO is many different medical providers come around, ask questions of the patient (many of them redundant), check the patient and then get together with the whole team to come up with a final assessment. It would seem that larger teams of specialists could be employed through the tool to achieve better outcomes.
Foreign Policy: The US frequently makes mistakes on the world stage simply because it doesn't understand the country or region it is interacting with. The Iraq war is a famous example of this. High level "experts" provide information and intelligence that people on the ground know is wrong. The information system we are contemplating can be used to break through this type of bias by including more voices, filtering in or out the correct people based on their expertise, and being able to properly debate policy positions before committing to them.
Consumer Services / Shopping: Most retail websites feature a rating system for the products they sell but it is unclear how honest these are. They are, supposedly, fairly easy to game with fake reviews, bot votes, etc. Our system will obviously be able to filter this out and be highly customizable based on product/service and who is doing the rating. It could replace product information services such as WireCutter (in the NY Times) or Consumer Reports which are presumably unbiased but not exactly clear about how their lack of bias is achieved.
Polling & Surveys: Polling is said to have become more unreliable in part because the individuals being polled are harder to reach or are skewing their answers due to their distaste for the media (which they consider polling to be a part of). For example, 2016 election pollsters failed to correct properly for the number of older white voters in key Midwestern states. Pollsters normally adjust their predictions from the raw sample data to reflect demographic reality (eg if the number of African American voters is under-represented in the sample, the pollster will adjust their weight upward in calculating the final result). However, polls can be done using our system very easily and avoid mistakes by getting feedback from more users. If a key demographic group isn't being counted correctly, users can come forward to make that clear. The algorithms themselves, in our system, can be adjusted to do the final calculations and their workings will be open to all.
Predictions: It is well known that if a large number of people try to guess the number of coins in a jar and we average their results, the final answer tends to be very accurate. This "wisdom of the crowd" effect has been applied to many problems and shown to frequently outperform experts. It relies heavily on independent thinking by the participants (they can't influence each other or be influenced by some authority), a diversity of perspectives (not all the same kind of people), and proper aggregation techniques. Our system will have or can be easily adapted to achieve these characteristics. Our network will be able to judge the independence and diversity of its sources, individually and as a whole. The aggregation techniques will fit in with the algorithms we are already developing for the network itself (Bayes, averaging, etc.)
Whistleblowing and Call-outs: Our system can be used to allow people to anonymously call out bad actors or whistleblow at organizations they are part of. The system would need an anonymous feature so individuals couldn't be identified but could be verified as real people (instead of bots). A recent example of this is [hospitals not revealing price information](https://www.wbur.org/onpoint/2024/02/13/the-fight-for-transparent-health-care-prices-in-america) for procedures even though it is the law that they do so. Either they have flouted the law completely or have complied in bad faith, releasing thousands of pages of cryptic procedure codes (along with prices), with no search functionality, that ordinary people will have difficulty interpreting. Those hospitals that have been targeted by complainers, and subsequently legal authorities, have quickly complied. Usually running a campaign to force this kind of change is difficult. However, since much of the infrastructure for information gathering, community activism, and consensus will exist in our system, it can be molded fairly easily to account for this situation.
Given the number and nature of the applications, it seems that an eventual extrapolation of this system would be to replace government (at least in its decision making capacity -- you still need firefighters) and governance in general. What is governance? Usually it is an elite group of decision makers, even in its democratic forms. But their elite nature, comparatively few numbers, and the political system they are part of practically ensures bias. Not to mention capture by interested parties and outright corruption. Polls (wisdom of the crowd) reveal a low trust level for politicians in general. Most countries have a few political parties which are dedicated to advancing a biased agenda (almost by definition). The US, the largest democracy and supposedly the vanguard democratic nation, has exactly two viable political parties, one of which is currently dysfunctional. And even at its best, US democracy is widely regarded as highly flawed.
The system we are building will enhance democracy by giving people more direct input, aggregating their results more accurately, and providing open debate which anyone can take part in. The community building features could be used to include anyone who wanted to take part in government, presumably from among the population that lives in a certain area. They can then "vote" through our system which will be designed (if they desire) to filter out bad actors, factual mistakes, and bias. Open and full debates on policy decisions can be made and the community can agree on how to tally up the opinions that get put in. If the decision is appropriate for the "wisdom of the crowds" then they can tune their algorithms accordingly. If it requires an expert, that can also be arranged. And, since governance will largely be treated as a voluntary effort by all, the system will be difficult to manipulate by vested interests.
This may seem like quite the vision but let's remember that the internet's original promise, that of a democratized information and collaboration system, has not really met its utopian expectations. We can certainly get information faster now but our social structures have not changed much. It would seem that the problem is not the internet itself, which is nothing more than a protocol and technology to implement it, but the software built on top. If the internet is simply an online newspaper rather than a paper one, that's progress, but it's not revolutionary. If it is online shopping, that too is progress but we used to be able to catalog-order things before the internet. If the internet is social media, well it's certainly nice to be able connect with friends so easily, but it has had disastrous side effects. The internet's original vision became dominated by narrow objectives and profit seeking. No one actually set out to build a software system which could organically change the way society functions, for the better. And these things don't build themselves.