Oarsman Pay Parable

Imagine an ancient oarsman, rowing in a galley boat. Rowing takes effort, and risks personal injury, so all else equal an oarsman would rather not row, or row only weakly. How can his boss induce effort?

One simple approach is to offer a very direct and immediate incentive. Use slaves as rowers, and have a boss watch them, whipping any who aren’t rowing as hard as sustainably possible. This actually didn’t happen much in the ancient world; galley slaves weren’t common until the 1500s. But the idea is simple. And of course the same system could also work with cash; usually make positive payments for work, but sometimes fine those you discover aren’t working hard enough. Of course the boss can’t watch everyone all the time. But with a big enough penalty when caught, it might work.

Now imagine that the boss can’t watch each individual oarsman, but can only see the overall speed of the ship. Now the entire crew must be punished together, all or none of them. The boss might try to improve the situation by empowering oarsmen to punish each other for not rowing hard enough, and that might help, but rowers would also use that power for other ends, creating costs.

An even worse case is where the boss can only see how long it takes for the boat to reach its destination. Here the boss might reward the crew for a short trip, and punish them for a long one, but a great many other random factors will influence the length of the trip. Why bother to work hard, if it makes little difference to your chance of reward or punishment?

There is a general principle here. As we add more noise to the measurement of relevant outcomes visible to the ultimate boss, the harder it is to use incentives tied to such outcomes to incentivize rowers. This is true regardless of the type of incentives used. Yes, the lower the worst outcome, and the higher the best outcome, that the boss can impose, the stronger incentives can be. But even the strongest possible incentives can fail when noise is high.

Yes, one can create layers of bosses, with the lowest bosses able to see specifics best. But it can be hard to give lower bosses good incentives, if higher bosses can’t see well.

Another problem is if the boss doesn’t know just how hard each oarsman is capable of rowing. In this case most oarsmen get some slack, so that they aren’t punished for not doing more than they can. This is just one example of an “information rent”. In general, such rents come from any work-relevant info that the worker has that the boss can’t see. If rowers need to synchronize their actions with each other or with waves or wind or time of day. If a ship captain needs to choose the ship’s route based info on weather and pirates. If a captain needs to treat different cargo differently in different conditions. If a captain need to make judgements about whether to wait longer in port for more cargo.

In general, when you want a worker to see some local condition, and then take an action that depends on that condition, you must pay some extra rent. So the more relevant info that workers get, the more choices they make, and the more that rides on those choices, the more workers gain in info rents.

A related issue is the scope for sabotage. Angry resentful workers can seek hidden ways to hurt their bosses and ventures. So the more hard-to-detect ways workers have to hurt things, the more bosses want to treat them well enough to avoid anger and resentment. Pained, sullen, or depressed workers can also hurt the mood of co-workers, suppliers, customers, and investors whom they contact. And the threat of pain can stress workers, making it harder for them to think clearly and well. These issues tend to argue against often using beatings and pain for motivation, even if such things allow stronger incentives by expanding the range of possible outcomes.

Overall, these issues are bigger for more “complex” work, i.e., for more cognitive work, work that adapts more to diverse and new local conditions, and work in larger organizations. In the modern world, jobs have been getting more complex in these ways, and the organization and work literature I’ve read suggests that finding good work incentives is a central problem in modern organizations, and that more complex work is a big reason why modern workplaces substitute broad incentives and good treatment for the detailed and harsh rules and monitoring more common in past eras.

The literature I’ve read on the economics of slavery also uses job complexity to explain the severity of treatment of slaves. Slaves in artisan jobs, in cities, and in households were treated better than field slaves, arguably because of job complexity. They were beaten less, and paid more, and might eventually buy their own freedom.

Bryan Caplan has argued that ems would be treated harshly as slaves:

Why wouldn’t ems’ creators use the threat of `physical hunger, exhaustion, pain, sickness, grime, hard labor, or sudden unexpected death’ to motivate the ems? Robin elsewhere talks about `torturing’ ems, so why not?” .. Modern systems of slave labor – see Stalinist Russia and Nazi Germany – used pain freely, because the penalty for quitting was death. (more)

When slave-owners have imperfect information about slaves’ productivity, high quotas lead to lots of counter-productive punishments. Threatening to execute everyone who falls below the 90th-percentile of output, for example, requires slave-owners to kill 90% of their slaves. Information about ems’ productivity, however, should be much more accurate, especially since most descend from a small number of exceptional humans. These are ideal conditions for heavy use of negative incentives. (more)

3. Moral and legal barriers aside, imperfect information about workers’ ability is the only self-interested reason not to treat them as slaves, especially when you can pre-select workers for docile personalities.
4. Since brain scans allow for cheap copying, employers would have excellent information about ems’ true abilities. Create a few dozen copies, give them life-or-death incentives to excel, and see what they accomplish. Then use that information against all the copies: Perform at your potential or we’ll inflict horrible pain on you. (more)

Caplan sees one particular kind of info rent falling with ems. And yes, with large em clans bosses would know more about the permanent general tendencies of ems. But individual ems would still differ due to different training and work experience. And more important, most info rents in modern complex jobs don’t come from info about permanent general tendencies! Even if you know a worker in great detail, you’ll still need to give them good incentives to react well to specific job context, and you’ll need to treat them well enough to avoid motivating hidden sabotage.
Most of Caplan’s disagreement with my book comes from his thinking ems are extremely docile. So extreme that (as I summarized):

Ems are not disturbed by “life events”, only by disappointing their masters. They only group, identify, and organize as commanded, not as they prefer or choose. They have no personality “in a human sense.” They never disagree with each other, and never need to make excuses for anything.

Yet the following is the only argument he’s given:

Only a tiny hand-picked sub-sample of humans will be copied. The obvious question: Why wouldn’t ems largely be copies of the most “robot-like” humans – humble workaholics with minimal personal life, content to selflessly and uncomplainingly serve their employers? (more)

Today workers who are taller have some advantages, like being able to reach taller shelves. But shorter workers also have advantages, like crawling in smaller tunnels. You can’t just look at one advantage of one feature and conclude that a feature is an advantage overall; you have to look at the net effect of that feature. Yes there are some advantages of employees who don’t talk back and never question orders. But there are obviously disadvantages as well. Sometimes you want workers to be creative, take the initiative, and to persuade and inspire others, and “robot-like” hardly seems best for these.

I’ve tried to look for data on overall effects:

Our best data should be the most productive people in the world today, or that we’ve seen in history. Yet the most productive people I know are not remotely “robot-like”, at least in the sense he describes above. Can Caplan name any specific workers, or groups, he knows that fit the bill? In writing the book I searched for literatures on work productivity, and used many dozens of articles on specific productivity correlates. But I never came across anything remotely claiming “robot-like” workers (or tortured slaves) to be the most productive in modern jobs. (more)

I didn’t find any data on slaves and docility, though I did find how docility fits into the standard five factor personality framework. Docility is lumped with “submissive, dependent, pliant” as part of “passivity”, which correlates most strongly and positively with neuroticism, but also positively with agreeableness and negatively with openness. In general only the agreeable part suggests more productivity in most jobs today; neurotic people are less productive, and the effect of openness depends more on job type. (more)

But Caplan hasn’t responded to these; “why wouldn’t” remains his only argument offered.

GD Star Rating
Tagged as: , ,
Trackback URL:
  • Pingback: Overcoming Bias : Caplan Debate Status()

  • Having a good idea of an em’s abilities isn’t much help if the noise comes from the difficulty of the task they are performing. Even a very good idea of an em’s ability doesn’t help much if you want it to solve different problems each time and there is substantial variance in the time to solve the same problem (as I expect there would be if ems were proving theorems or writing code).

    Of course, one could probably measure how much effort an em was putting in fairly easily. Not only could you make sure the em was particularly poor at faking hard work and gather data on the em’s tells you could simply examine the simulation of it’s brain to see if the appropriate brain regions were being used.

    However, the part of Caplan’s argument that seems totally unfounded is the assumption that extreme and stressful punishments would be the most effective way to extract the most work from ems even assuming their performance could be measured.

    If ems are like most humans I would expect the stress and worry induced by such terrible threats to make them worse at complex intellectual tasks. On the other hand if they are robot-like workaholics it would be superfluous to threaten them as they will have no inclination to slack or do anything besides put their all into doing their assigned task.

  • Alfred Differ

    There is also the problem of acquiring productivity gain. One might come up with a decent strategy for maximizing productivity with
    slave-like employees, but productivity gain is needed for competing groups. Try doing that with docile employees and see how far you get.

  • lump1

    I think that neither you nor Brian are realistically modeling the precarious situation of the em who must pay the market price for CPU cycles with wages. Analogies to productive workers in our economy are beside the point, because they are not facing death if they don’t economically outcompete all their rivals. But this is exactly the situation all ems are in. Forced laborers in WW2 are a much better analogy, but they still had it easy compared to ems, because they only need to satisfy their guard in order to survive. Ems need to out-earn *copies of all of their most productive rivals* in order to survive, a much taller order.

    We could all earn an extra dollar if it meant the difference between our life and death, and another, and another… but if the price for breathable air kept going up so that even the most productive workers would struggle to afford it, our approach to our jobs and lives would change profoundly. The survivors wouldn’t be anyone’s slaves with whips at their backs. They would be the kind of people willing to sacrifice everything simply to be able to breathe another day – their leisure, their non-existential priorities, their dignity and their scruples. A less complete commitment would mean that the available air will be sold to someone else who earned more, and can pay more. I’m trying to say that this argument about slavery is completely a red herring, because em scenarios are in no way analogous. A much better analogy is that em life is a permanent lifeboat crisis, where an effectively infinite number of strong swimmers are competing for a finite number of rafts. Forever.

    • Dave Lindbergh

      How is that different from the way all wild animals have always lived?

      Isn’t this just a description of the “state of nature”?

      • Not the actual human state of nature. Humans aren’t cut out for the “state of nature.” Could society even exist in such a state?

      • Pretty much all humans before a hundred years ago lived at near subsistence level incomes.

      • What I meant (and thought David Lindbergh meant) by “state of nature” (inasmuch as it’s in quotes and also fits the context) is the state of all against all – strict egoistic competition. I think we’re crossing Hobbes and Malthus.

    • A less complete commitment would mean that the available air will be sold to someone else who earned more, and can pay more.

      But since ems (and humans) don’t work effectively under such stresses, wouldn’t it be in the employers’ interest to shield their ems from such dire competition.

    • Alfred Differ

      I don’t get why the price of cycles would climb that way. Suppose it did and produced ems who behaved that way. Why should we believe that will impact productivity in a positive way? What about productivity gain?

  • kurt9

    I saw someone on Caplan’s blog mention the old materialism vs. dualism argument. Dualism, except for the narrowly defined information paradigm used for discussion of “uploading”, has been utterly discredited as a viable concept for well over a century.

  • Riothamus

    I am skeptical of the enslavement and torture arguments. Why would it make sense to go to the trouble of torturing them when for every environment we could just choose to copy the most successful ems *in* that environment?

    I also suggest there would be diminishing returns. In order for torture to be effective, I infer the ems would have to devote compute cycles to remembering and anticipating the torture. In regular humans, too much suffering induces trauma, which is where our memories and expectations interfere with our lives.

    I expect that there would be curves of compute cycles devoted to remembering/anticipating torture, and also of productive compute cycles. I propose every em would have a ‘trauma point’ beyond which the added compute cycles devoted to torture exceed the added productivity cycles, thus *reducing* productivity. Even supposing torture occurred, it seems much more efficient to produce many copies of ems who are at the trauma point already, eliminating the need for further torture.

    When the labor supply is functionally infinite and the cost of firing and hiring both approach zero, why even bother finding out how productive torture would be?

  • Elias Hakansson

    In “Good to Great” Jim Collins’ team of researchers found that so called “level 5 leadership” was the single most advantageous asset a company could have in improving a business’ prospects of success. Level 5 leadership is characterized by a paradoxical mix between personal humility and strong professional ambitions. The personal humility exhibited by these personalities would manifest itself in the way they forfeited celebrity status and admiration from others and channeled that effort into ensuring that the company did well.

    I don’t know how a docile slave would willingly internalize the objectives of his workplace and ensure its long-term success if he has a hostile relationship with his boss and his workplace. Investing in motivating personnel is a largely fruitless effort (motivational speakers, team building etc). According to Collins et al workplaces wherein employees are internally motivated outperform those where they are externally motivated. Internal motivation is another aspect of Level 5 leadership.

    Granted, this only appears to apply to leadership positions and creative positions, but in an Em world it is conceivable that AI is sufficiently advanced to complete almost all mundane tasks. In this case the only positions worth enslaving Ems over would be held by AI, and for all positions where level 5 leadership is required Ems would be used.

    So I don’t think slavery would occur in an Em economy.