模倣子 Bus Bingo

 Memetics Index 


I've been working as a bus aide on the special education bus in my school district. This means that I help the driver to get children seated and keep order, basically. Together with the driver, we decide the seating chart for the whole bus. The issue of concern that I addressed in this project was that of students not getting out to get on the bus in a timely manner, resulting in delay in getting to school on time.

A bit of background. The special ed (SpEd) bus route consists of "curbside-to-curbside" delivery, that is, students are picked up individually, and dropped off at their target school, usually to be met by a school aide. Each pickup has a scheduled time, and arrival at a school is coordinated to correspond to when the aides will be waiting on the curb, when the school opens, and things like free breakfast at the school. Unfortunately, students not getting on the bus on time, even just a few of the total, rapidly degenerates into late school arrival and things like being late for class and getting tardy slips and missing breakfast.

Obviously, if students recognize that they share responsibility for the whole bus arriving on time and behave accordingly, none of this would be a problem, but they do not. It dawned on me that this was effectively a macromemetic engineering problem, i.e., getting a group of people to behave collectively in a selected fashion, i.e., getting out to the curb on time, and engineering it in terms of a shared set of memetic behaviors and iconic tools. The question is: which memetic behaviors and icons?

As with most all memetic engineering projects, the task is to undermine or subvert apathy. The general principle here is to engineer immunomemes ("bullying opportunities" (1) ) which allow students to transmit social approval or disapproval to their fellow students based on the to-be-optimized parameter, in the case, timely ridership, and make this a high and consistent priority for all of them.

A further issue in memetic design is "injection" (4). As the bus aide, an authority figure, I was able to hand out items ("ballots") to my riders and have them accept them, for the most part, and I could offer my ability to change seating arrangements for them (5). They were in turn motivated to ask me the meaning of the ballots I had given them, without my having to take the initiative, indeed, those that learned the system were in a position to transmit the information to others, which happened to some degree.

I did a memetic analysis, drawing the needed diagrams and listing memes I thought were needed for the system. After some reflection, I came upon an implementation tool: the Bus Bingo Ballot.

The Initial Analysis

I drew up some initial state transition diagrams. I envisioned a bunch of verbal memes like "good job being on time" or "you're late!" which so far I have not yet implemented or seen emerge organically (15). This may be the fault of not having completed my analysis and design. Having said that, in The Blue Shirt Tuesday Doughnut Day system, the verbal approval and disapproval memes developed organically and I really just chronicled them after the fact.

fig. 1.1. Initial System Design notes

The "carrot" I hit upon was to give students who "behaved well" the ability to choose where they wanted to sit on the bus and with whom. As the bus aide, this is one thing I control. Humans love freedom and control, and being told what to do on the bus is a great source of resentment among riders, and so I reasoned that giving riders this freedom of choice might be a strong positive motivator. Given the requirement of fulfilling my primary mission as aide on the bus, and the limitations of the bus environment, I was very constrained as to things that I was able to make the students respond to, which is, of course, a problem of apathy (and injection (4)). Whatever tool I devised had to be easy to understand and use while on the bus.

My further design explorations led me to the idea that students should be in three macromemetic states (9): seat at risk, seat safe, and able to choose seat, or SeatRisk, SeatSafe, and Chooser. Here's some of the analysis I did for that.
fig. 1.2. Initial State Transition Diagram Sketch

The idea is that if a student shows up on time they go into and remain in a SeatSafe state, and if they play my game (which is what became "Bus Bingo") then they can get into Chooser state (assuming they are in SeatSafe already), and if they fail to ride on time, they are in SeatRisk state. Those in SeatRisk are susceptible to losing their seat, getting moved somewhere not of their choosing, by students in the Chooser state. If in a SeatSafe state, they cannot choose, but, in principle, cannot lose their seat, or even, possibly, not have another student they don't want moved into their seat with them (2).

I had envisaged further states wherein submitted bingo tickets would be eligible for a lottery drawing of some type, giving random access to seat choices, but I soon dropped that idea.

The Initial Requirements

I needed a simple tool that would be easy for me to produce and update. I needed to be able to somehow mark which students had participated in the daily Bus Bingo. Obviously students needed to be able to see who was a bad or good rider, and who had won the right to choose their seat.

I had initially thought of a kind of lottery, where I would choose at random from the tickets and give that person the right to choose their seat. I quickly discarded that idea in favor of just letting everybody who was an on-time rider AND turned in a "ticket" to me that same day (every day to maintain it) be in the "Chooser" state, able to make seating recommendations to me.

Here's the device I came up with, which so far is working without much in the way of modification. It meets the requirements of the results of my analysis in terms of showing the states of the system and  memes allowed to the riders. 

fig. 2. Bus Bingo Ballot (blurred)

I've blurred it to hide the students' names. In a big bus implementation, presumably full names would have to be used. There is a circle on every name and empty seat space. A blue circle means that the student's seat is safe from being changed by somebody else, and that the student is able to ask the aide for seating changes. A green circle means the student is an on-time rider (and has caused no other disruptions) and that their seat is safe. A red circle means that the student has been late or disruptive and the seat is up for grabs. The student can be moved at will by the aide or by any blue-circle students. Blue circle students may also discuss among themselves to make new seat assignments and bring them to the aide.

This is simply a letter size piece of paper folded three ways, and run through a color copier, and I used colored pencils and a Sharpie to make it. To mark which student's ballot it is, I make tears on both sides of the student's name so that it can fold up like a tab. "Voting" for another student is done by making a tear sideways through the student's name.

The ballot shows where all students sit, which seats are in at-risk, which are good riders, and which are empowered to decide the seating arrangements. I have tabs at the top for the aide and the bus driver, which I use to mark my own ballot during the route for when students are late (or misbehave).

A couple of goals which I have yet to achieve, and which I presume will be needed for a more autonomous "big bus" implementation are to get the students to mark their own ballots, both with their name, and voting for other students. A future big bus implementation might not require students to self-identify, but merely to vote for their fellow students as either "bad" or "let them off the hook," so this might be simpler. My SpEd bus implementation offers the ability to choose one's own seat, so it's necessary for the ballots to be self-identified, while on a long bus, anonymous ballots might serve.

A further early objective which I quickly abandoned was where students could mark their seating preferences on the ballot. I decided it would be easier to just talk with them, and this turned out to work well. The idea of students writing on the ballots for this and other objectives appears to be untenable (10).

Memetic Pairing, Marking, and Closure 

This is a critical concept to memetic design. Up to the time of starting this project, I had told students that they needed to be on time in order for me to consider their requests to be reseated next to friends and so on. This was in fact a frequent conversation. My requests for timeliness in exchange for seating changes were seldom effective, certainly not in the long term.

There are a number of macromemetic problems with what I was doing before. One is that "be on time" and "I'll give you the seat you want" are not very good memes. They lack the macromemetic properties of marking and closure, or rather they have "poor" marking and closure (12). This problem is solved by the tool of the Bus Bingo Ballot. The BBB solves the marking and closure problems associated with behaviors like boarding the bus on time, not misbehaving while on the bus, and being ready to get off the bus at one's stop. Whatever happens, if you do these things right, you get a green circle, if you fail, you get a red one. Since it's on the BBB, all the other riders are able to see the results of your behavior. The difference between the aide nagging you every day and having your behavior reflected on a tool like the BBB comes down to marking and closure issues. The latter is more real and permanent and is visible to all others, so it's easier to link to other memes.

Another critical concern in memetic design is memetic pairing. Simply put, behaviors are tied to results, often those provided by gatekeepers and authority figures. For example, "give alms, go to heaven," or "go to work, get paid," and so forth. This seems pretty obvious, but in the case of the SpEd bus, there is no such quid pro quo linkage in effect. I paired desired behaviors to certain results, and bad behaviors to other results, in this case, getting a green or blue circle on the ballot, or a red one, and then I further paired those memetic icons to additional memes that the students were able to deploy. To wit, getting a blue circle meant you could choose the seating, red meant you could not, and indeed could get kicked out of your seat, and green meant you were safe, could stay in place.

I introduced memetic states and clear memes which mitigated the transition between these states. For example, here are the "deployment descriptors" (13) for the system.

SeatRisk.late! => SeatRisk
SeatRisk.on-time! => SeatSafe
SeatSafe.submit-ballot! => Chooser
also SeatRisk.on-time!submit-ballot! => Chooser
[Chooser, SeatSafe].late! => SeatRisk

fig. 3. Deployment Descriptors for Bus Bingo system

Effectively the Bus Bingo Ballot makes the states clear for all to see, which in turn makes clear which meme deployments are available to each rider. The states and new memes (which are designed to maximize closure and marking) are things which I, as the memetic engineer, invented from whole cloth, except for the desired meme, i.e., on-time! I created all the other states and memes as scaffolding to support the increased deployment of this on-time! meme by my riders. But the reality is that late! and on-time! are subjective behaviors which I had to clarify as part of the system design. 

What constitutes being on time and late? My goal was to have the students define this themselves and mark and vote each others' names accordingly, and the same for good/bad behavior, but I have not been able to get this to work. However, this does not mean that the students do not believe that this is what is happening behind the scenes, that is, that I, or their fellow students, or some murky combination, are marking the ballots somehow. The fact that the system works as well as it does I attribute to the "Effigy Effect," which I discuss below.

So in a sense I have created all of the memes involved in the above deployment descriptors, since I not only created the Bus Bingo tool, but I also defined what late! and on-time! mean, hence defining those memes which sort of already existed.

So I paired the getting of the right color circle to getting to choose your seat. I also paired boarding on time with getting the needed circle. So in a sense, I decoupled the boarding on time behavior from the asking the aide about seating behavior, which was not working. This is the meat and potatoes of memetic engineering--clarifying memes in terms of marking and closure, and pairing them together in ways that also have good marking and closure, and defining states between which these memes cause transitions. As I did in the Prime Pizza Thursday project, pairing a desired meme which has unavoidably bad closure and/or marking with one that has good closure and marking can solve the problem of getting a cohort to reliably deploy the desired behavior. "Boarding the bus on time" may in fact be that sort of mushy meme, and pairing it with circles on a Bus Bingo Ballot may solve that problem.

The "Effigy Effect"

This macromemetic principle merits special consideration. One obvious evidence of this was the "rebel" (3) students expressing concern (2) that they lose their seats or have a student they didn't approve of getting assigned to their seat.

I did not overcome apathy in all of the ways I had hoped, e.g., voting, offering verbal encouragements, asking about the lateness of students who had boarded before so they could vote them on their own ballots after the fact (a form of "snitching" (11)).

The idea is that behind some macromemetic interactions there is an unseen "Big Other" that is influencing the process. So even though there are no visible memetic deployments from visible fellow agents, such as encouragement, recognition of good ridership, or chastisement for being late or other bad behavior, the effigy effect allows riders to imagine somebody watching, unseen, and deploying memes such as giving one a red circle or such. This takes the onus off of me, as the aide, since I have the excuse that "some people must have noticed you were late yesterday" or "somebody probably saw your bad behavior" or even that any of the blue-circle students are planning to take your seat away (even if none actually are).

Even though the project effectively "failed" in terms of engaging students to overcome their apathy around voting and even marking their own names on their own ballots, they accepted the results of the ballots they were handed every day, e.g., the colors of the circles changing, as to an extent the "will of society" rather than just the aide doing things all by himself, keeping his own counsel. In a sense, the rule-based nature of the system telegraphed this idea (14). Also, the very visible blue circle and the fact that students with a blue circle could dictate the seating assignments was a very palpable message. In fact, students were motivated to get a blue circle even if they had no apparent interest in changing seats.

Summary & Conclusions 

The Bus Bingo project almost immediately produced good results. Students who were chronically late became good riders, i.e., showing up on the curb at the pickup time or up to five minutes before. The bus was no longer waiting on students. We started showing up early to our student transfer rendezvous, sometimes so early that we moved on to our second rendezvous, whereas before our rendezvous bus was leaving before us. Of course, getting to schools on time became workaday.

My goal was met, i.e., fixing our timely ridership and schedule problems. I was not, however, able to get students to vote, certainly not reliably, to put their fellow students onto the "naughty list," or take them off of it. The Effigy Effect seemed to cover this, however. Even though I was the one changing the states on the Bus Bingo Ballot, there was the credible belief that other students were at least partly, and facelessly, contributing to transitions from SafeSeat to RiskSeat, and to Chooser. It was, however, outwardly visible that some students were choosing their own seats, and this was a clear incentive, and students were actually encouraging each other to get to the bus in a timely manner so that they could choose their seats. This was rousingly successful.

One obvious future target is the "long busses" or the non-SpEd busses. These pick up multiple students at single sites, i.e., no "curb-to-curb" and importantly, they don't have bus aides. This means that the system must be more self-operating, and there are a larger number of students involved. This means an administrative load to process the system. Also, rather than timely boarding of the bus, "discipline" is probably the major target behavioral problem. The bus driver is solely responsible for driving the bus as well as keeping the students in their seats and curtailing other undesirable behavior. My problems with apathy associated with voting are key here. The non-SpEd riders will have to reliably vote each other as "bad" or "good" ("off the island," so to speak) and have this translate directly and relatively seamlessly into seat assignments, e.g., "bad" kids have to sit up front.

In short, the Bus Bingo has been enormously successful almost immediately, and without a lot of modification or tweaking, in achieving its stated aims. It failed in engaging participants in terms of marking their own ballots, that is, replying with detailed information, however, which will be an issue in future long bus implementations. The Effigy Effect seems to have been powerful, since riders reacted strongly to the movements in the ballot system, even though they themselves were not "voting" the changes, nor was anybody they knew. More to the point, the system created the desired sense of community, even though riders were relatively apathetic at interacting with it, apart from making the effort to show up on time and behave well. This case of the Effigy Effect may be an example of yet another previously unobserved effect of the Second Law of Immunomemetics (14), i.e., the outward appearance of the functioning of a system of rules creates the assumption that immunomemes are operating behind the scenes, even though none are. In other words, the appearance of fairness (or democracy) may be taken at face value—yet another interesting moral lesson from macroeconomics.

Future Work 

1. Simplifying and clarifying the current BBB implementation (if possible).

1.1. Implement BBB on a spreadsheet to reduce admin time, increase adoptability

1.2. Write a super-simple, super-clear recipe book on how to implement BBB for use by other SpEd buses in other school districts

2. Devise a version of BBB usable on non-SpEd buses ("long buses")

2.1. Minimize admin time commitments by using spreadsheets, etc.

2.2. Get a voting system that works so students can regulate their behavior as a group

3. Explore the idea that bribe-driven memeplexes may not be able to be self-sustaining or to mutate (6,7,8)

4. Investigate the relationship between the effigy effect and the 2nd Law of Immunomemetics


(1) Obviously the macromemetic term "bullying opportunities" is problematic in a K-12 school environment. A stand-in term I thought of is "interaction opportunities" even though this fails to capture the idea that students are sending messages of disapproval (or approval) to their fellow students, based on whether or not they show up for the bus on time.

(2) One thing I observed immediately upon implementing the Bus Bingo System was that "non-participating" students approached me with concerns that if they were good riders (SeatSafe), they be able to approve any student getting moved into their seat (in this case, students who initially had a seat to themselves).

(3) See this Blue Shirt Tuesday essay. A "rebel" is one who opts out of participation in a memetic system, but who nonetheless adheres to all its tenets (as opposed to a "criminal" who breaks the rules for some personal gain). A rebel may be thought of as a kind of "non-participating proselytizer."

(4) Injection is the process of inserting a collection of memes, a system of memetic behaviors, a memeplex, into the members of a population (a "memetic cohort") such that the memeplex may begin to function. This is effectively the "catalytic barrier" of macromemetics. A successful marketing campaign is an example of injection, making people aware, convincing them to use your memes (one of the most important action memes for a marketing campaign is "buy my product").

(5) Submitting to the will of a trusted authority figure, so to speak, or gaining the attention of one, can be classed as a "libidinal reward" or an act of "libidinal investment." Offering a reward, or simply, a "bribe," makes it easier to inject a memetic system (6).

(6) This was the tricky bit in my Box Binning project (still need to write up) where there was no reward associated with compliance apart from the bullying opportunities against coworkers for non-compliance, which were being injected at the same time. Ultimately this worked and the Box Binning system became self-sustaining and ultimately highly contagious and free to mutate--all highly desirable qualities in a memeplex that sustains corporate profits and ensures process reliability. It's possible that memeplexes that require the continual injection of resources or of the work of specific individuals (7) cannot be self-sustaining or freely mutating--this is perhaps an interesting area for further study.

(7) The Blue Shirt Tuesday experiment obviously required the weekly infusion of free doughnuts (8). This offer of free food, a bribe, is a motivator for engagement with the system, for injection. An interesting variation might be the addition of memes where the bringing of doughnuts might be shared. This might be possible, but it was never attempted. In this Bus Bingo Ballot system, the bus aide (or driver) makes and hands out the ballots and interacts with riders to update the seating arrangements. Doughnuts, attention of authority figures, control of the environment (in this case, where one sits) are all bribes, elements of libidinal investment on the part of the inured agents (riders who have been injected with the Bus Bingo memeplex).

(8) And the Prime Pizza Thursday Project offered the bribe of free pizza for participation.

(9) Defining the states of a memetic system is central to the overall memeplex design. See the Three Laws of Macromemetics. Meme deployment is what actuates state change. For example, a student deploying the meme of showing up late results in either staying in the SeatRisk state, or if in the Chooser or SeatSafe states, transitioning to the SeatRisk state. Likewise, a student deploying the meme of on-time! goes to the SeatSafe state, and if they submit-ballot! they go to the Chooser state. Hence state design is right up there with meme design in putting together a memetic system, and like any other critical design phase, getting it wrong causes problems later on.

(10) Having the students write detailed information on the ballots seems an unworkable complication, and there's no guarantee that all students will have things to write with, and sharp pencils and such on the bus can be a hazard. In order to proceed to phase two, a "long bus" implementation to deal with discipline problems on the regular buses, simplifying and systematizing things down further and away from the more complex is the way to go.

(11) "Snitching" or enforcing the rules of the system even when it does not directly serve oneself, is crucial to the proper functioning of a memetic system. Unless members enforce the rules, those rules which are not enforced cease to be rules (see the Laws of Immunomemetics).

(12) See the glossary for definitions of "marking" and "closure." These refer to whether the meme is clearly recognizable, e.g., did the rider board the bus "on time," for marking, and "did the aide give the rider the seat they wanted?" for an example of closure. Another way to think of closure is "can all agents involved see that the memetic transaction was completed or not?" Marking is about "can all agents recognize the meme?" Both of these interactions score poorly on both of these critical properties. However, "You've got a blue circle on the ballot," which has high marking and closure both, and can then be paired with "you get to choose your seat." Another issue is whether the whole cohort, all of the other riders in the population, are able to see the status change. In other words, other riders might miss the fact that a given rider was late, or is consistently late--especially if they board after the rider in question--but they can always see their status on the ballot regardless of any of those muddying considerations. This is another example of good marking. Likewise, giving a rider a green or blue circle for certain behaviors (boarding on time, turning in the Bus Bingo Ballot regularly) gives those behaviors good marking, and also good closure. Vague verbal contracts fail to do this for many reasons.

(13) A deployment descriptor depicts how deployment of a meme results in the transition from one system state to another. For example, Hungry.eat! => Sated. A set of deployment descriptors is the most "complete" way to describe a memetic system, that is, a state transition diagram, while easier for a layperson to understand, rapidly becomes unwieldy for large systems. Furthermore, it may be more difficult to spot "return failures" (situations where one can get into a given state but not back out again) or other design problems with deployment descriptors, and also to implement certain design tricks such as "Packing the Meme Space."

(14) This may be an interesting example of the Second Law of Immunomemetics, i.e., that any system of rules must be accompanied by a set of "bullying opportunities" (an immunomemeplex). In other words, if there is a stated rule such as "don't be a dick" or "cheating will not be tolerated," but in fact there are no actions which may be taken to prevent people acting like dicks, and cheating is rampant and nobody is doing anything about it, then these "rules" are really just so much dross, so much feel-good nonsense. Contrariwise, in the case of Bus Bingo, there may in fact be no immunomemes that are actually being deployed, e.g., students voting against each other for bad behavior, but since there are stated rules, e.g., "board on time or get a red circle" and this is actually happening, i.e., late students are getting red circles and good students are getting green and blue ones, and blue circle students are getting to choose their seats, all in accordance with the "stated rules of the system." This is an interesting result of the Second Law, i.e., that "fairness" breeds "compliance," and that the perception of democracy, even if there's little reality behind it, is satisfactory and works.

(15) Memes which I had hoped to "engineer" like "was so-and-so late?" asked by students who had boarded earlier and wanted to mark their own ballots have begun to appear organically, however, so this is an encouraging mid-course emergent empirical result.

No comments:

Post a Comment