模倣子 Prime Pizza Thursday

I wanted to conduct an experiment where there was a sporadic offer of a reward, or rather, a sporadic (but predictable) appearance of a MIAO. The MIAO in question was free pizza, with the requirement that one supply:

1. a prime number
2. an "idea" (1)

The way the experiment was conducted was that when a Thursday fell on a prime number, free pizza would be provided, and a clearly-labeled box explaining the rules and a pile of pens and paper chits were placed next to the pizza. For instance, it would be one large pie, or one large pie on each of two floors, or two on one or both, each with an "idea box." The rules were simply that one had to to write a prime number (3, see below) on a chit along with "an idea."

One objective of the experiment was to elicit the thoughts and opinions of the cohort (2). How better to get people to answer than free pizza? Also, to ally the anxiety about whether one had to give a "good" or "right" answer or idea, the requirement was primarily to provide a prime number. This made it tacitly clear that the "idea" was secondary, and could really be anything. This was re-enforced by sending out the results to everyone every week in an e-mail. Everybody could see the kinds of responses which were acceptable, both in terms of how silly and also how serious, critical of the company, etc.

A further objective of the experiment was to see if cheating would be suppressed by the rule (meme) system, as we found in the Blue Shirt Tuesday Experiment (10), i.e., no cheating, almost perfect conformance. Would people cheat for pizza where they had not for doughnuts, and so forth.

One thing I found was that there was no cheating.  I got about as many responses as slices of pizza, if not more.  This was a more accurate result than Blue Shirt Tuesday (BST), i.e., I knew how many slices of pizza were provided, and I received a box full of chits with ideas. This was all anonymous.  
After a year or two was that a consensus was developed around the fact that a large number of people wanted better vending machines, some content specified (certain flavors of Monster energy drinks were lacking, apparently), and that the vending machines should accept credit cards.

Again, no cheating confirmed, and could be proven (pizza slices versus number of responses). Demonstrated that it was possible to get a cohort to engage in a kind of "voting" process, where they might otherwise be apathetic to other appeals to "voice your opinion" and "you can control what goes on here."  As a matter of fact, I didn't take any action to do anything with the responses until I started to get questions along the lines of "what are  you going to do with these answers?"  As I recall (11), I thereafter began asking specific questions, e.g., "do you want credit card vending machines" and such.

I also got commentary outside of the PPT about some of the "ideas," e.g., I heard some opposition to the "sei-ri-bi 生理日" (12) suggestion, along the lines of "whoever keeps putting in that monthly menstruation sick day only for women, I don't think it's fair, what a jerk."

I did not notice any global effect of cheating producing new memes to prevent the cheating, which I also observed, albeit less rigorously, i.e., did people without blue shirts take doughnuts -- who knows? One question might be how to measure this?

Since there was no cheating in either PPT or BST, I was not able to observe whether the memetic system adapted in response to cheating, e.g, "tougher" memes allowing cohort members to bully cheaters, etc. This sort of experiment (13) could be very interesting in terms of at least a few points:

1. do memeplexes react to these kinds of environmental factors?
2. can memeplexes adapt to address "problems"? (14)
3. would there be more cheating in a more heterogeneous community? Why?
4. this has widespread social implications in terms of diversity, violence, crime, peacekeeping, etc.

Another interesting phenomenon is how the specification of a very simple and trivially easy action to satisfy the qualification for free pizza, i.e., write down a number (an easy to determine one, even if it's just the current date (6). This may be a trick that leads to high compliance and participation. Some government agencies do this, notably the US Tax Office (the Internal Revenue Service), i.e., just filing a form by the specified time qualifies the filer for "compliant" status, regardless of whether the information on the form later prove to be incomplete or incorrect. Other forms for correcting the information are made readily available. Making compliance difficult, and penalties for failure to comply menacing may lead to less compliance, and a greater need to pursue wrong-doers and administer penalties. If the way is easy and the burden is light, one may be able to count on social bullying almost completely, and for the cohort members to do all the work of getting the compliance information when needed (later).

A further strengthening point is that since the results are published (anonymously), with the prime numbers submitted, and then the "ideas" going with each, cohort members get further reinforcement about what is "acceptable," leading to a higher likelihood of compliance. This should be true since if I, as a cohort member, can see the types and "levels" of answers are acceptable, I can tailor my own answers to fit into the bound of this norm.

One interesting question is the "perception of others cheating." I was working with a community  of engineers, so it was perhaps obvious to this cohort that the number of responses sent out each Prime Pizza Thursday (PPT) should be equal to or greater than the number of slices of pizza distributed, i.e., typically 16 to 24. If the number of responses (4) were less than the slice count, then the whole population would know that cheating was taking place. What would be the impact on the memes of the cohort? The theory of memes suggests that the effect would not necessarily be that of generating more memes to bully cheating (14).

This may have been the experiment where I started to see the idea of residual memetic debt  (15) for the first time. It was certainly true for the "Koffee Klatch" (7) experiment conducted later. As with Blue Shirt Tuesday (BST) (8), I found that cheating was practically non-existent.

The evincement of the response, i.e., filling in opinions was accomplished by three or more factors: 1) anonymity, 2) reward of free pizza, 3) social pressure of memetic system, 4) other. It's unclear which were preëminent.

I verified conclusively that cheating was low to nil in the population I selected. Again, the injection of a simple engineered (artificial) memetic system proved to be very easily and quickly done. I demonstrated that it was also easy to evince a very specific behavior (9) from the population which was was next to impossible otherwise.

I did not really get back any useful information in terms of sporadic stimulation of a MIAO. There was no interesting difference that I could determine with respect to BST. This may not be true, however, in fact, but I have not yet analyzed it to tease any such information out.


(1) the "idea" was very loosely defined, in fact not defined at all.  Some took the form of suggestions for changes to the company, many of them spurious, e.g., deck chairs and free beer on the roof, etc. Basically, one was asked to write something, which could be anything once the prime number was supplied.

(2) with the understanding that 90% or more of the responses would be spurious or irrelevant.

(3) people had a lot of fun with this, e.g., some prime number were expressed as algorithms, complex formulae, in binary, or interesting prime numbers such as the smallest prime number which may be written as a palindrome, etc. Some people were lazy and just wrote 5 or 7, however.

(4) I may decide to post all of the results, but it will be quite a large collection of files, so this might be prohibitively disruptive. I may try a trick such as posting them "in the past" and then bringing them back to life afterwards. This experiment took place over a period of two years or more, and the number of PPTs was (I need to check this) zero to two per month, so there would only be a few dozen result sets.

(5) this would "help" cohort members, i.e., ensure that PPT would not be terminated due to too much cheating (if that were in fact the effect that cheating would have...)

(6) which was almost never done, by the way, except for the 5th, oddly enough -- something to do with Star Wars. "Red Five standing by...?" Writing the ideas in red pen was also suggested.

(7) this was a rather briefer experiment, conducted later. I have some other articles relating to residual memetic debt (RMD) which I'll try to link here.

(8) See essay(s) on BST (10). This was where free doughnuts were laid out, every Tuesday, on condition that one be wearing a blue shirt. It was testing a couple of things: 1) whether the behavior of a cohort could be manipulated and how easily given a simple reward, and 2) whether social bullying could be effective in curbing "cheating" and how effectively. The results of both of these seem to "very easily" and "very effectively." The cohort was a bunch of engineers, so it would be interesting to try against other (perhaps more varied) populations. It was an experiment in memetic injection, and it demonstrated conditions under which it may be accomplished. Again, more diverse populations may change the playing field.

(9) which experience showed was very difficult to evince voluntarily, i.e., the written opinions of the cohort about anything. People had a lot to say, a lot to complain about, and a lot of suggestions to make, yet they did not make these suggestions. Part of this may have been the anonymity provided. One openly took the pizza, but one's responses were not attributed (including, notably the "monthly monthly use-it-or-lose-it sick day exclusively for women" idea).
(10) In  What Do Memetic States Look Like? I take BST as an example for the graphical notation. In Cheaters, Masochists, Sociopaths, and Proselytizers I look at how memetic systems put people into roles, or niches, and how that shapes society. In Practical Memetic Engineering I compare BST, PPT, and Koffee Klatch and examine the approach to getting people to behave in a designated way. In Libidinal Bribes and Memetic Debt and Choosing Libidinal Bribes I discuss things like how physical rewards can produce compliance and how the nature of those rewards and how they are dispensed can affect "cheating." In Anthropomorphicization and Cults of Personality in Social Conformity I look at how memetic systems (memeplexes) enable specific bullying behaviors which may (or may not) defend the memeplex and keep it running in stable and "good" way.
(11) I will consider posting the rather voluminous collection of responses over the approximately two-year period. Please "Like" or comment on this essay if you'd like to see them.
(12) A "sei-ri-bi" (生理日) or "menstruation sick day" exists in Japan. A woman is allowed to ring into work for one day every month for a "free," "use-it-or-lose-it" sick day. These special sick days don't add up if not used. In practice, some women take more time off, and at least in some cases, this is allowed. I proposed this frequently as an idea in the box, i.e., that woman at our company be given these special sick days. It was not re-submitted by anybody else, and it was assumed that one of the American women had submitted it (and it was put down to the one American woman engineer, since all the other female engineers were foreigners...who for some reason were assumed not to have suggested such a thing...I guess).
(13) for instance, one could imagine "agents" assigned to cheat as part of the experiment, or setting up the pizza or doughnut rewards so that it "looks" like cheating were happening (goodies disappearing more rapidly than expected or such), or even just propagate false messages that cheating was occurring. Would the memeplex adapt organically to this, and would the adaptations be "successful" in curbing the cheating, make it worse, or product other results (coherent or otherwise)?
(14) In theory they cannot, i.e., memeplexes please themselves, and they do not perceive rampant cheating as a "problem" in the same was the humans "governing" the cohort would.
(15) I discuss residual memetic debt in Contact Memes and the Corporation, The Tragedy of the Koffee Klatch, Libidinal Bribes and Memetic Debt, and Practical Memetic Engineering. I also have a definition for it (which may be outdated) in my Memetic Glossary.
Memetics Essay 模倣子

No comments:

Post a Comment