Productivity is a key measure of economic progress, typically defined as output per worker per hour. In the US, productivity has increased about 5 times since 1947, according to inflation-adjusted data from the Bureau of Labor Statistics. This significant increase aligns with observable improvements in living standards over that period, including advances in food availability, medical care, technology, education, and overall comfort and convenience.
Measuring productivity in a ratings-based society could involve both subjective ratings and objective metrics. One approach would be to calculate a baseline productivity for an organization, then adjust individual productivity based on hours worked and subjective ratings. A normalization factor would ensure the sum of individual productivities matches total productivity.
In a moneyless society, numerical values could still be assigned to production for statistical purposes, without using them as a market pricing mechanism. Goods and services would be distributed based on need and ratings rather than price. A hybrid approach could use money only for luxury items while providing basic needs without monetary exchange.
Effective debate should focus on identifying the Pareto frontier of possible solutions, then choosing points along that frontier. This approach helps get to the crux of arguments and enables productive discussion of trade-offs, even for contentious issues. New information may shift the Pareto frontier, but repetitive arguments on settled issues should be discouraged.
Trust can be viewed as the probability of favorable outcomes in interactions with someone. It is based on prior experiences and develops over multiple interactions. A robust ratings system should differentiate between inherent trustworthiness and uncertainty due to limited information. Context and culture influence trust, often favoring large institutions. A well-designed ratings system could help balance this, enabling more distributed trust networks in communities. 🤖
[//]:#(End AI-generated summary section.)
<h3>Productivity</h3>
Productivity is normally defined as output per worker per hour. It is the central measure of economic progress in any society and the basis for increased standard of living (at least in material terms). The following graph shows how US productivity has increased since 1947, the first year the Bureau of Labor Statistics started tracking it.
To the question of whether this is based on real GDP (ie inflation adjusted) or not, the answer is yes. GDP over years of time is normally plotted after adjusting for inflation since inflation would be too much of a confounding factor (our inflation since 1947 is a factor of 13). According to the [source above](https://tradingeconomics.com/united-states/productivity), "Labor productivity is calculated by dividing an index of real output by an index of hours worked of all persons..."
As we can see, productivity in the US is about 5 times higher today than it was in 1947. The first thing to ask about numbers like these is whether they make sense. Do they pass a basic smell test? This is often a critique of US vs. European standards of living. Proponents of US style laissez faire capitalism will point to the social democracies in Europe and declare that we are doing much better than they are based on GDP per capita. But a passing visit to any of the countries in question reveals that their people live just fine. Some things are better, some worse, but on balance middle class people in both areas are at a rough parity.
So would a subjective dive into your likely 1947 life make you conclude that your standard of living was 5 times lower then? This is hard to quantify but it seems within the ballpark. There was substantially less food, medical knowledge was drastically lower (no antibiotics), entertainment/information was limited to radio, print publications, and the occasional movie (in a theater), you may have had a telephone in your house (which everyone shared), educational opportunities for most were limited to high school, at best (the GI bill was just getting started which made college a viable option for the masses). The level of human services was also diminished. Mom’s pretty much raised their kids themselves. Airplanes were not yet a mainstream mode of travel. Cars were not as ubiquitous and more people used public transit (probably not such a bad thing). More people lived where they grew up. More people did physical labor for their jobs. Buildings were not air conditioned or heated as well. Life was quite a bit harder and more uncomfortable. But we still had many of the benefits of modern life. A factor of 5 seems reasonable.
How would we measure productivity in a ratings-based society? Obviously we could have a rating for productivity which would give us a subjective view of each person’s contributions. But we would probably want a more direct objective measure as well. Although the factor of 5 seems reasonable, it’s not at all clear that we would have arrived at that prior to seeing the numbers.
In a moneyed society we would use the value of what everyone produces. This is hard to break down to an individual level but for a company, for instance, it might result in dividing its revenue by its employees and then differentiating between the employees in some manner that adds up to the total. The differentiation would be subjective for employees that don’t have clear output metrics (eg managers) but could be objective for those who do (eg line workers). In an economic system with no companies, we could perform this same exercise at the community level.
Here we create a baseline productivity ($P_b$) by taking the total production of the firm ($TP$) and dividing it by the total hours worked ($THW$).
$P_b = {TP \over THW}$
If we then assign this number to each individual and multiply by their hours worked we would obtain the total production. $H_i$ here is hours worked for each individual:
We will assume that hours worked is objectively measurable through clock-in/out procedures or timesheets. If we then multiply this by a subjective productivity rating for each individual ($R_i$), we obtain a total production which is less than the actual total production (because the ratings are 0-1). We call it the total baseline production ($TP_b$):
So we introduce a normalization factor ($f$) which is simply the actual total production divided by this number.
$f={TP \over TP_b}$
This factor ensures that the sum of individual productivities always adds up to the actual total productivity:
$TP = f \displaystyle\sum_{i=1}^n {P_b} {R_i} {H_i}$
In a moneyless society, we could do something similar if we were willing to take the step of assigning numerical values to production. This would be subjective but keep in mind that subjective judgements of value are ultimately how the moneyed economy works too. People make judgements, subjective ones, and assign prices to things based on them. For instance, the price of luxury cars would drop to the price people were willing to pay if we all decided they weren’t worth that much. A moneyless value assignation scheme is only different in that we wouldn’t also have a <i>market</i> pricing mechanism.
But the subjective notion of value in a moneyless economy is certainly a step in the creation of money and shows just how natural the concept of money is. But we could remain determined not to introduce money and simply use the productivity numbers for statistical purposes. This is useful, even with no formal market mechanism behind it.
If there is no market-basis for value, how would we assign prices? The answer is that we wouldn’t. In a moneyless society goods and services are distributed based on need first and ratings second, as [we have discussed](Brainstorming_6). A pricing mechanism is basically a voting system and our proposed economy would simply not have that particular one.
By the way, we could also do this in a moneyed economy. There is no economic law that says pricing is determined by what people are willing to pay. Goods and services would exist and would be distributed according to need and ratings. We could simply assign prices to the goods depending on how difficult they were to manufacture (ie computers are more valuable than can openers) and use those prices to calculate productivity. If someone were to order a new product and win the claim to it, they could simply be given the money to “buy” it. Or the money transfer, such as it were, could be handled automatically from the “bank” to the firm. Again, we may choose to have money to perform statistical tasks but not use it as a scarce resource or a price voting mechanism.
We could also choose a middle term approach where money is not used for most things but is used for luxury items. Since we advocate a system in which everyone receives a minimum basket of needs, and we assume the post-scarcity condition of an industrialized western nation, this basket can just be handed out to people. There wouldn’t need to be any money. This is akin to the way certain services are already provided, such as public K-12 education. Beyond that, we would distribute most other goods based on ratings. But we could have a category of scarce luxury items that do require money, in order to ascertain their price and match them with demand. We wouldn’t want our society to devote itself inordinately to the manufacture of luxury items but it may be motivational for some people to have access to them. The money, therefore, can be distributed based on ratings and limited based on how many luxury items the community wanted to produce (presumably, not much).
Note that in most cases cited here, we have rejected money and market-based price judgements. We emphasize that this does not detract from our ability to measure productivity and use it as an important objective function in societal optimization.
<h3>Debate and the Pareto Frontier</h3>
Debate should reduce to an exercise in choosing along the Pareto frontier. This is to say that without first establishing the Pareto frontier, a final debate involving policy choices should not begin because it will more than likely involve dominated alternatives. Once the Pareto curve is understood, the debate can proceed to the somewhat more subjective matter of choosing a point on it. For instance, a community may be choosing between supercomputer designs and has a Pareto front with Performance vs. Cost. The choice boils down to how much they want to pay for each level of Performance. The stage has thus been set for a productive debate.
This is an example of getting to the crux of the argument, an important property that debate should bring out or, in this case, start with. All debate should ultimately try to find the Pareto frontier and then choose from the points along its curve.
This idea should apply even for more abstract and contentious issues, such as abortion. The abortion debate often revolves around when life begins. The pro-life movement believes it begins at conception and the pro-choice movement believes it begins, for practical purposes, quite a bit later. However, almost no one in the pro-choice movement thinks that infanticide, right after natural birth, is permissible. Similarly, they don’t believe in abortion the day before a baby would be born naturally. They believe in some range, well after conception and well before birth. The decision of when to draw the line is an example of a tussle along a Pareto curve with only one dimension, the time in pregnancy beyond which it would be illegal/immoral to have an abortion. The pro-life advocate wants that time set to the lowest possible number, and the pro-choice advocate the highest possible number without going over some limit. Once the debate reaches this point, settling it becomes a subjective judgement or simply a practical compromise dictated by political considerations. Further debate is pointless, without new evidence, since we have succeeded in finding the “Pareto curve”, simplistic though it may be, and have identified the crux of the argument.
Note that the abortion debate relies heavily on our knowledge of the fetus. If a fetus is found to have surprising cognitive abilities, for instance, the pro-choice advocate would probably agree to move the permissible abortion time to an earlier point. This means that changes are possible to a settled debate but only with new information. Incidentally, this is one reason the pro-life movement has been so successful of late. Through better imaging of the fetus they have found medical evidence suggesting that fetuses develop critical functions (like a heartbeat) earlier in pregnancy than was once thought.
Communities, along these lines, might stipulate that once a policy is enacted post-debate, it cannot be changed without substantial new information. That is, debaters are not allowed to continue arguing settled issues. Since there will be no law (hopefully) preventing debate in these circumstances, ratings themselves will have to perform the work of shutting down repetitive arguments and argumentative people. An unrated sandbox will always exist where people can argue for its own sake. One idea might then be to take unending debates and relegate them back to the [sandbox](Brainstorming_13) and stop providing them with new levels.
The abortion debate can be more sophisticated, of course, than simply the time at which life begins. Let’s suppose that, as a result of the debate above, we have passed a law that outlaws abortion after 18 weeks of pregnancy.
The debate can now move on. We could introduce a two-dimensional objective function by considering policies that result in more or less abortions. Presumably we would want to reduce the number of women who seek abortions in the first place, whether legally or not. A policy that gives women support for childcare, for instance, might be put forth on the grounds that it will remove the financial incentive to have an abortion. We can contrast that with a policy that, say, steps up enforcement of the abortion ban after 18 weeks. So, we have two policies and two outcomes. Both of these would have their own cost-performance Pareto curve which we might plot as follows.
Here the childcare option dominates the law enforcement option and should be the selected policy. There is simply no mechanism of law enforcement that is more effective at all cost points. Keep in mind here that the law enforcement policy is, by itself, Pareto optimal. That is, it is already the optimal policy in terms of how we should design it (tactics, manpower, budget, etc.). Therefore no one can argue that we haven’t considered some aspect of law enforcement that improves the situation.
Here we have a cost where there is crossover. One policy dominates below a certain cost and another one above. The community will probably have budgetary constraints anyway, so the appropriate policy can be decided easily. If not we have, again, a subjective judgement to make. But at least we’ve identified the bone of contention.
It is likely that the pro-choice debater will want to argue against the law enforcement policy no matter what because they are against any abortion restriction. But we already have a settled law on the books banning abortion after 18 weeks. The pro-choice debater cannot re-argue that issue without new knowledge. We stipulate, furthermore, that they cannot enter into the new policy debate in good faith by relitigating the old one.
What can be done, however, is to introduce a second independent variable for Pareto consideration. In addition to cost, we might have privacy intrusiveness as a variable to be minimized. This amounts to a new insight, new information if you will. We would argue that, all things being equal, it is better to opt for a policy that minimizes the state’s intrusiveness into someone’s life. Certainly in this case, the law enforcement policy doesn’t look quite as good. We could plot this new variable on a third axis of the graph but then it starts getting hard to capture visually, [as we’ve discussed](Brainstorming_11). A better option might be to combine it with the cost variable. In other words, we’d be looking at cost in terms of the community budget and also in terms of privacy intrusion. This essentially moves the law enforcement cost curve to the right:
With this curve in mind, the community is prepared to make a decision.
Nearly all contentious issues can be reduced in this way as long as the debate stays focused on extracting the crux of the argument and not getting mired in tangential issues. Once the crux is identified, simulation can help produce the actual Pareto frontier curves for use in decision-making. The ultimate purpose of debate is to do this, not provide a forum for rancorous argument or entertainment.
This debate could also veer off into philosophy by diving into the notion of when a fetus should be considered a conscious person, with the legal protections that all such persons enjoy. The pro-life argument is usually religious and precludes any speculation in this direction. But the pro-choice side, by limiting abortion to well before the natural time of birth, might have some ideas about this. Even the pro-life side could jump in, sensing an opportunity to move the upper limit to an earlier point. Such a debate is permissible, of course, but it should be removed from a policymaking arena. It is likely to be inconclusive, as many philosophical debates are, and have an impact only in the longer term.
It is also likely to be held by academics who have advanced training in philosophy. This is not to say that we shouldn’t have such debates. They, like scientific debates, further our understanding of the world. It is philosophical and scientific inquiry that has informed our thinking on matters of animal consciousness, for instance, and shifted our views toward their more humane treatment.
Let’s take another traditional debate that appears to be ideologically driven but is really an optimization problem: how much the government should spend on the poor. Before we start, let’s keep in mind that any argument that invokes personal circumstances should be rejected outright. No one can say “I don’t want my tax dollars to help the poor” or “I am poor so I want that program”. Personal circumstances arguments should be rated by the system as unworthy. People can certainly feel that way, and express that sentiment, but not in a debate that is trying to extract valuable public policy guidance. This, btw, is how Rawls would have it as well, limiting debate to publicly accepted basic principles. There are exceptions to this idea, of course, such as when an individual circumstance serves to introduce valuable public information, eg “the policy, if enacted, would result in my death because I wouldn’t be able to afford medical care”. Generally speaking, however, the personal complaint is not an argument.
We are left with an economics and morality argument. What is the amount of help a society should give the poor to maximize economic performance (or HDI)? At which point does the help become immoral because it provides too little? At which point does it become excessive? These points again, can be reduced to objective functions and analyzed mathematically.
This particular debate is important because it touches on Rawlsian issues of basic liberties. If we do not have enough food, we are not really free to participate, especially as equals, in the workings of society. We aren’t free to pursue an education or the job we want (both Rawlsian rights). So the economics debate about helping the poor is only an exercise in optimization once the moral dimension of the question is clear. If we have decided unequivocally, as a society, that we will not permit starvation then the optimization question becomes a matter of how best to accomplish this. We are no longer debating whether or not to permit it. In a sense, the moral debate, like the “when life begins debate”, is a one-dimensional Pareto curve but this time with a binary choice: whether to permit starvation or not.
A binary-choice debate, such as this, is relatively easy to resolve. A community will debate the matter and will either persuade all its members one way or the other, or create a split opinion. A split opinion would then lead to a vote and a decision rendered. This is straightforward enough with the exception that such a vote (or debate) might be precluded by some sort of higher law (constitutional) which has already rendered a decision. In such a case, we presume the debate is already settled. If members want to change the constitution, the vote would have to be conducted in accord with such an objective (eg require a super-majority).
We might conclude this section by commenting on the rigor advocated here for how debate should be managed. A critic might react by saying that free people should be able to debate however they like. There shouldn’t be rules about how they are conducted or how they end. We shouldn’t be required to optimize an objective function when all we’re trying to do is talk to each other. This is fair criticism. But every polity needs its rules of order. No government can function if everyone gets to speak at once. In a direct democracy, the kind we are envisioning, this is especially true. Of course, people will be allowed to have unstructured debate. The debate system itself will even have a sandbox where unrated debate can take place. But the mechanisms by which policy is decided needs to have some kind of organized approach. The creation of a flexible debate structure, where everyone can participate, but leading to optimal outcomes, should bridge the gap between an overly restrictive system and an unproductive cacophony of arguments.
<h3>Thoughts on Trust</h3>
Trust can be looked upon as the probability of an interaction with someone having an expected favorable outcome. This might mean they tell the truth when asked a question. Or it might mean they don’t cheat you in a financial transaction. The outcome must be expected because normally we don’t count on random outcomes, even if they are favorable. And the outcome must be favorable because that is, presumably, the reason for the interaction in the first place. More generally, trust in a person can be viewed as the overall probability, taken over many interactions, that future outcomes will be favorable.
Trust is based on prior interactions and its accuracy depends heavily on sample size. A single favorable interaction with someone does not confer the same level of trust as 100 such interactions. Our trust level should converge to some value after many repeated interactions with someone.
The following graph illustrates this for a case where someone’s average level of trust is 0.7. Here, each interaction is rated for favorability, between 0 and 1, and the cumulative average favorability is set to 0.7 (our trust level). Indeed, it takes roughly 80 interactions to reach a settled-out view of their trust, although we’d have a fairly good idea after about 30 interactions.
Trust is, of course, more complex than a simple random analysis like this would suggest. Many people are more consistent in their interactions such that, in this case, each interaction would deviate more tightly around the 0.7 than this graph suggests. In that case, it would require fewer interactions to establish a reliable average trust.
But trust has many more complex aspects to it. We often have to calculate trust on the fly with people we don’t know very well. We resort to heuristics and trust by association. We hire people if they have the background we want and “seem ok” in a single interview. We trust the bank teller to honestly handle our money even if we’ve never met her before. And, needless to say, even the most rigorous calculation is subject to statistical error. Over a period of time, we continue to adjust our number up or down depending on the objective outcome of various interactions.
It is important to understand how trust is derived and to separate out the various components that make it up. This enables us to precisely differentiate between our problems (inability to predict) and someone else’s problem (choosing to do wrong). When we rate other people, we want to rate them fairly on the basis of their inherent properties (eg morality) and not on our own inability to predict them because we don’t know them that well. How often have you answered a question about someone by saying “well, they seem nice but maybe they’re an axe murderer at night”? A good ratings system should cleanly distinguish between the methodologies being used.
It should also account for another human characteristic, the tendency to give people the benefit of the doubt. If we don’t know someone, we generally assign them a baseline level of trust. We do not automatically assume that people are untrustworthy because we don’t know them. This too is a recognition that our knowledge of someone is frequently lacking, but we know that it is probably ok to deal with them in some limited way. Our ratings system should probe for the reasons for trust levels and categorize them appropriately. Fortunately, by integrating over large numbers of people (ie interactions) the ratings system enables this possibility.
Context and culture obviously matter a lot in trust. If a stranger on the street asks you for a large loan that they will pay you interest for, you would probably be unwilling to trust that person. But we are willing to walk into a bank and deposit our money with equally complete strangers. The fact of the bank is important. But so is the fact that it is normal to walk into a bank and hand them money for deposit. It is not normal to do so with a stranger.
Note how this asymmetry of trust greatly favors the centralized institution. This is rational up to a point. The bank, for instance, is not likely to “run away” with your money but the individual might. The bank can, furthermore, spread its risk so even if it does something stupid with your money it will be able to cover your deposit using someone else’s. And the bank is presumably already trusted because it already has many customers. It wouldn’t exist otherwise. But the monopoly of trust enjoyed by the bank seems to go too far. There is no “banking industry” at a small-scale individual level, offering higher interest rates (let’s say) for the objectively higher risk its customers would be taking.
The same is true for many institutions that, to put in pointedly, hoard trust: insurance companies, hospitals, companies that sell consumer products. Size matters in large part because trust rises in proportion to it. The law frequently magnifies this trend with naturally favorable treatment of organizations that have the heft to comply with its regulatory requirements (and have political influence).
This has two negative effects. The first is that we have learned to rely on institutions rather than each other, thus weakening community bonds. The second is that the institutions themselves gain too much power. It’s a difficult problem to get out of once a culture has accepted it.
A functioning ratings system is obviously the antidote to all this. If the bank has a built-in ratings system, because of its customers, the individual would have the same with our ratings system. Someone with high financial trust could offer, for instance, personalized insurance services at lower cost because they know you better. Or banking services at higher interest rates. Meanwhile the ratings system will also be accurately gauging the trust assigned to institutions, to make sure it is deserved and not just an assumed value based on size, advertising, or some other irrelevant factor. Again, this brings to mind the importance of ensuring that we know exactly why people trust as they do.
