Back in March, Zack Kanter wrote,
Amazon has opened itself up to inevitable ‘gaming’ by sellers. Another way of saying this: as soon as a system’s rules are understood, it will be gamed according to the rules that have been created.
Pointer from Tyler Cowen.
He alleges that Amazon is allowing sellers to buy or manipulate their way into top spots in Amazon’s rankings. There is much more in Kanter’s essay, but I will just focus on the issue of gaming the system.
Any system invites gaming. If you run a sales force and give bonuses based on a formula, your sales people will game the system. If you reward volume, they will make concessions on price. If you reward sales revenue this year, they will invest less in building long-term relationships. If it costs you more to service some customers than others, then unless you can figure out how to build that into the sales formula, they will sell more to the high-maintenance accounts.
Government regulations invite gaming. I have often used risk-based capital requirements for banks as an example.
Search engines invite gaming. The phrase “search engine optimization” still grates on my nerves.
All rating systems invite gaming. Don’t think that we won’t see bots writing book reviews and product reviews, if that isn’t happening already.
There are many stories and jokes about gaming under central planning. If the planners reward the number of nails manufactured, they get a lot of little nails. If they reward the weight of nails manufactured, they get one giant nail.
Systems for compensating doctors will be gamed. If you reward them for seeing many patients, then they won’t spend time on the hard cases. If you reward them for doing certain procedures, then they will do more of those procedures. If you reward them based on outcomes, they will select patients that they know will recover easily. Etc.
If you have ever wondered why your organization rolls out a new compensation system every few years, gaming is the reason. When a system is new, it usually takes employees a while to figure out how to game it. After it has been around awhile, they have become experts at getting the maximum reward for the minimum effort. So the organization revamps its compensation system to try to induce more constructive effort and less gaming.
This is very true. The only qualifier I’d add is that mature systems tend to be more susceptible. In the early high-growth phase of a relatively new system, it is easier to extract value by participating in the system than it is by gaming it.
Great post. You seem to be hinting that gaming is a fundamental reason why markets outperform government command and control. I would love to see you take this farther.
I was thinking about this too — I think a market is also a “system”, it’s just that the rules for “gaming” it as a producer/supplier are (usually) to provide value.
The general reason that markets outperform centralization, including govt command and control, is because markets harness distributed, localized information. One type of distributed information is information about how to game a system: what’s cheaper, desirable, hard/easy to do, etc. Markets incorporate that information into prices, which removes the profitability of gaming. If there are too many small nails, the relative price of small nails falls making large nails relatively more profitable. Vice versa if there are too many large nails.
Good point.
= – = – = – =
I was reading Taleb’s book _Skin in the game_ and he said something about “citation rings” in academia. I believe that is the term. If you have one person saying a crazy thing, that person is crazy. But get a couple of dozen people saying the same thing and citing each other, keeping a journal going, that can be part of a career strategy and it’s a new perspective in scholarship. Independent of external demand, too.
The market would note that most people don’t want the output of the citation ring, and its product has little value.
This analysis is imperfect. The Greek and Latin classics don’t have a lot of market demand, but I consider them valuable anyhow.
My understanding of math is that that progress in math takes place independent of the usefulness of new findings.
Milton Friedman once wrote that things like Classics could be supported by private foundation endowments. Maybe he’s right.
In general the gaming / vs. markets is a useful approach. Hope this makes sense–I’m thinking on the fly.
Arnold, an old saying “en español”
https://cvc.cervantes.es/lengua/refranero/ficha.aspx?Par=59406&Lng=0
Tom Schelling used to talk about this in his classes, especially in getting his kids to do what he wanted. Often he couldn’t.
I knew that any system would have problems, so my wife and I didn’t have many rules for our kids. We let them know (partly by example) what we thought were good ways to act, and then trusted they would basically behave as good human beings, because they wanted to (of course, we hoped their vision of “good human being” was the same as our’s in the ways that mattered). They did some stupid things, but basically it worked.
Any successful organization has to have a fair amount of unwritten rules/esprit de corps/organizational culture. So does any successful society.
So you taught them Virtue Ethics as opposed to Deontology? Good parents.
One of my favorite papers, on a taxonomy of tools for addressing the problem of gaming: https://mpra.ub.uni-muenchen.de/90649/1/MPRA_paper_90649.pdf
https://www.illinoispolicy.org/chicago-defends-title-of-highest-sales-tax-rate-in-the-nation/
Chicagoans pay the city 1.25%, the county 1.75%, the transit authority 1% and the state 6.25% – for a total sales tax rate of 10.25%. The Tax Foundation specifically noted that many Chicago-area residents will shop online or in the suburbs to avoid paying the city’s high taxes. Even that work-around is being disrupted as online market places will be required to collect the state’s sales tax beginning Jan. 1, 2020.
—-
How about a reshipping service? Client gets a mailing address in a low tax state where they have their on line purchases sent. Then the re-shipper sends it to the home address in Chicago.
You left out the obvious conclusion: the system will eventually learn to game its users back. This is why all systems sooner or later fail, or become intolerable.
These are great examples of unintended consequences. It reminds me of work-to-rule strikes some labor unions employ. All of these cases emphasize just how much trust underpins day-to-day transactions.
Of course, the real problem is that we cannot quantify what we “really” want and rely on imperfect proxies for measurement. Analytically, it looks like a form of monopoly/market/bargaining power with better specification/higher correlation between proxy and goal analogous to greater competition.
Part of the solution may simply be to get rid of the game altogether whenever possible.
Take education. There’s no reason you really need to have a system of test scores or that sort of thing to determine teacher compensation (if you were allowed to actually do pay for performance in education). I remember back in high school that it wasn’t difficult at all, for either students or parents, to determine who the stellar teachers were, the average ones, and the awful ones. The principal simply ranks each person based on their knowledge of what they are doing from a holistic perspective and pays people based on that.
Goodhart’s Law.