Fact Machine: a 2010 Census Diary

          It was my intention to write from my temporary job this year as a census enumerator from the very beginning. Somehow, I never got any of it written down while it was still fresh in my mind. I'll try to do this narratively, and hopefully that will jog my memory about the procedural and theoretical curiosities. Much like the times I've worked for the Board of Elections, my intention is to emphasize the virtues of bureaucracy, or rather, how a well-organized system of the mostly-ignorant can both generate accurate information and protect individual rights.

          I'm aware that many commentaries are beyond critical of blunders, evils and abuses. Many of these are typical small-scale workplace cynicisms or highlight the gap between planning and execution. I consider both criticism and the things criticised symptoms of a working social measurement machine. On the whole, census employees themselves are cynical and practical if they're not overly eager. I am most critical of the amount that my positive spin is an outcome of training myself to persuade people to respond. It also seems to me that the public news about this has been superficial, uninformative, and passive aggressive, so I also see this as an opportunity to discuss attitudes at a personal level.

          About a month into the job, employees were sent a notice about what they reveal in social media, emphasizing that they must be clear that their opinions are not those of the Census Bureau. Also, of course, reminding them of their confidentiality oath, which I'll get to. I'm not sure whether particular incidents resulted in the reminder. I did tweet briefly while in training with another trainee in Louisiana, and I heard enough to know that the relative sanity and ease of my local operation was the exception to the rule.

          I took the test last summer (quite proudly one of the first to finish it), and got a call a few weeks later saying my score was in the 98th percentile. Good, but easily beaten by anyone with the 5% extra given to veterans. They called me at least twice again asking if I were still available. My response was always yes, but not until the third call did they offer an actual job. Contrary to the expectations I had that the census would require far more office than field work, it is during May's NRFU (Non-Response Follow-Up, pronounced "narfoo" - oh how the government loves its acronyms), a.k.a. door to door interviewing, that the bureau's payroll bloats by a factor of about 4. (I got this information from a news report I now can't find, my deepest apologies.)


          Interestingly, coverage of the census has focused more on the jobs it created and money they government spends on it than on what it actually is. It's interesting to note that, partially because I had scheduled a ten day trip to Germany at the peak of NRFU, I've only made twice as much actually working as I did during the 40 hour training week.

          Only in retrospect, knowing that, do I consider that training week to be rather too long. It seems they hired as many to be trained as possible, expecting to eliminate many of the trainees, and the hiring bloated so much that it was nearly impossible for instructors, themselves only trained the previous week, to even communicate enough with the next level up to get materials and organize field work. It was at this point that I referred to the organizational structure as being nothing like the already-redundant root system of clerks and supervisors provided in our handbooks, but like a katamari ball. Extension of this analogy becomes ridiculous.

          The training location wasn't set until fairly late, and I think our instructor may have been the last to know. Training was in a cram school in Flushing, a 30-minute bus or subway ride from the neighborhood we all lived in and would work in, and the transportation costs were paid. The sardonic instructor, crew leader for half the class, set the attitudes of the more socially attentive of the group. The different supervisor for my area was even more down-to earth. For a highly social job there seemed to be a high percentage of the less socially attentive, so much so that I felt gregarious in comparison. It is, after all, a job for those shaken loose from other jobs, and did not require an interview. I gained about 10 pounds that week thanks to Flushing's pan-Asian deliciousness.

          One major joke that came up was the fact that the government thought to issue us all ballpoint pens for those carbon-paper forms that required them, but blue ones. The fingerprinting required of all Federal employees required black pen. Hilarity, we all carefully agreed. Fingerprinting is regarded as a fine art involving careful choreography. Because the only people available to do the fingerprinting had had a single day of training the previous week, a lot of the fingerprinting had to be redone in the office later in the week with a touch-sensitive machine. Whether the amount of extra time paid to trainees amounted to enough to have bought enough of these machines for all the classes, I don't know. It only occurred to me later that it was especially important to fingerprint census enumerators because the temptation to do a little breaking and entering is strong. It's the momentum of the thing.

          We swore the Oath on the first day. We put our hands over our hearts and swore... jeez, I'll have to look this up. I'm not being paid to write about this, and I made sure to check that I was allowed to leak instructions to enumerators and such (all very boring). The central tenet was that we could not use any of the personal information we gathered for any purpose other than the census. I understand it may still be hard to trust someone based on their taking an oath (I left the "so help me God" part out, even), but there's a quarter million fine threat behind that oath, plus jail time. And again, I'm aware incentives are better than threats - more than aware, after my door-to-door experiences - but we all both took that pledge seriously and thought it a goofy cinch. More on that later.

          Although it at first seemed tedious to go through the many practice questionnaires covering different ways we'd have to fill out the form, the only one I never had to use was the instance of an address that was a vanished trailer on a yard. Even then, I had to do many "deletes" for joined, mislabeled, or under construction homes. I don't know what it is about any classroom situation that turns everyone into a surly teenager uneasy about how they appear to other students, but the reluctance to actually pretend to be performing interviews was stupid. After filling in time with practice, paperwork, and a test one of whose official answers was wrong, we did go "into the field", and discovered that most of it was the repeated writing of Notice of Visit forms. A scourge, as it must be noted as a personal visit, of which we were only allowed three.

         Technically we were all supposed to start the day after training, but the mid-managing level understandably delayed that at least two days. Trainees in areas requiring even more enumerators were just left in not-enough-supervisors limbo for a while.

         I would like to emphasize that I consider this kind of obvious human frailty combined with persistence of the structure - everyone knew where they were supposed to be getting what, and didn't just go rogue with their badges counting whoever they thought needed counting out of zealousy or institutional mistrust. Discussing problems wasn't penalized, but making problems official was mostly understood to be counterproductive. As an apparatchik, I simultaneously understood myself to be paper's tool, and also knew that paper needed me to be smart and responsible and make empathetic and logical decisions. It's the lack of imagination, the discussion of alternatives without initiative, that keeps the information gathered consistent, and the information-making apparatus predictable. Predictability is not only critical when we see this as demographic science, but also maintains the trust of the people being measured. The census is an extensive self-check whose only goal is thoroughness, above all else.


          There was another day of field training under supervision, a meeting of the whole local crew, and then, by the middle of the first week of May, we had all been issued binders for one or two blocks of buildings and the addresses of everyone in those buildings, including the names of those who had already responded by April 15. We could not have another binder until we had completed every questionnaire for the non-responding addresses. We were on our own.

          The immediate obsession of most enumerators is the dreaded Refusal. In that first week there were a fair number of people who'd simply forgotten, neglected or sent in late their mail-in questionnaires. But those that require in-person visits are naturally far more likely to be suspicious and reluctant, and literally the second door I knocked on was an RE. I developed a pleasant persistentness I doubt I could maintain for door-to-door sales - and naturally, since most people are more used to the idea of strangers ringing their doorbells selling something, the first priority was making sure they know that's not what I was. The second was making them understand how little information I wanted and how bound I was not to share it, and sometimes I never got that far. I have a particular personality that's very concerned with the sharing of information, but also feared that this eagerness would be counter-productive. It's an extremely emotionally draining job, with soaring highs and crushing lows and everything in between.

          As reformed in Title 13, the census is deliberately intended to be independent of other data collection, especially the IRS and INS. Not that my saying so will reassure immigrants. All of them must come from somewhere with some form of population estimation, for the determination of geographic vote allotment or otherwise. But the principle of keeping all authorities' hands from knowing what the other hands are doing as a means of protection may be counter-intuitive even - maybe especially - to those raised in the American educational system. And then again, why trust me just because I trust the principle? Bureaucracy too easily is considered to go hand in hand with corruption. The same law that criminalizes my leaking anything within a boundary of paperwork also criminalizes not cooperating with me. Census policy amongst enumerators is to keep most ignorant of this threat, partially because prosecuting is so much harder than allowing a wider statistical margin of error.

          While I have a massively bothersome conscience, it's not that I can't imagine how the census system could be abused. I know that the census materials, seemingly completely innocuous, could be used for identity theft if leaked. What's on the short-form questionnaire can't be use for that, of course, but with a few tweaks you could get social security numbers. It probably wasn't in my interest, but when people used the excuse of having "done it over the phone" or "already sent it back", I was willing to suggest that they hadn't been responding to the real census, especially when we'd finish and they'd ask "that's it?"

          It wasn't particularly useful to take on an air of authority or even dress as well as they wanted us to. I enjoyed the position of authority too much to be utterly casual enough to be asking birthdays innocuously, but approaching that was far more effective than confrontation. Of course, there are many people who respond badly to both chattiness and officiality. Fast=painless was a useful thing to mention when not going by the book, the book being just going straight through the questionnaire.

          One critical limitation was that we ourselves were not a proxy respondent. Information could not be generated by the observations of the enumerators, only by the responses of the interviewee to questions. I find this a very important argument against the all-too-easy The Truth Is Out There model, in which facts already exist before they are recorded. The census form itself carefully delineates the grammatology of facts - who or what counts as a person, and a residence, and an occupant, not to mention gender and race, and the especially fuzzy Hispanic Origin distinction. Any change in the shape these values fit into changes the fact.

          Bruno Latour is the source for this kind of post-de-structuralist understanding of the social mechanisms of knowledge. Even looking at the root of "fact", you can see that it's something that is made, arduously. Information does not want to be free. Information wants to be wrong, and countering that entropy demands labor, and anywhere there is labor there is the question of pay.

          It may be easier to think about hard sciences as a collaboration between human, non-human, and hybrid objects when we look at the humanist biases of the census's statistical baseline for social science. Every quantum of personal information intends to be self-identification. Since it's faster to just let an informed respondent (who must be over 14) describe the entire household, this is merely an ideal. Critically, as with the Pirahã language, everything requires a source. At the NRFU stage, this is even limited to only a person that a person working as an enumerator has talked to. There is something very interesting that happens to social interaction when it is ordered by paper and print infrastructure as a means of constructing facts. The strategies of enumerators often aim to downplay this weirdness despite the legal carefulness of official instruction.

The very distinction between social and physical is limned by the human-paper work collaboration called the Census. By limn, I mean that it is a hybrid that creates the boundary by entangling its sides as much as possible. That we are specifically gathering social facts is entangled with way the bureau has defined society as made of humans. Like their lag on computer-based forms, this is one aspect of the census that I want to say makes it outdated in comparison to almost all other statistical sources - but embedded in that criticism is a contradictory The Truth Is Out There Error, as the census makes the understanding of society as without people.


I tried to build up a reputation of near-superhuman efficiency. I estimated it would be possible to actually complete two forms an hour, if it took 10 minutes for an interview and 5 minutes to fill out an NV (Notice of Visit) form. We were told that the main thing they were able to note about our performance was the number of completions per hour, but naturally we could tweak this based on what hours we reported.

The absolute rule was No Overtime: if we filed for more than 40 hours we would be immediately fired. This was never a problem because it was so unlikely we would find anyone at home and answering the door at anything other than the evening that it was pointless working entire days. Of course, we were encouraged to vary our hours, so that we weren't always visiting a particular home at the same time every time and wasting our three in-person visits. Especially once I came back from Europe and was essentially mopping up complete non-responders, we often had to work weekdays in order to interview superintendents and building managers during their working hours.

I had a binder with a list of addresses and the EQs (Enumerator Questionnaires) for two buildings still waiting to be finished when I got back. I'd done most of one building and none of the other. While I'd been gone, extra paperwork (a list of submitted cases for each day) had been added on to try and improve the way EQs were being sort of lost between enumerator and area supervisor and prevent overworking crew leaders. Naturally, we would forget to mark the status of cases in the binder before turning these in and then wouldn't remember what to write in there. I'm not sure how the address lists were used, but they were the source of the most sensitive bulk name/address information.

The mop-up we were doing at the end of May was mostly EQs without binders, anyway. I became the person sent out to clean up after census takers who had already quit or been fired, usually quietly. I am quite proud of being good at this job, and do wish there were something similar I could do permanently. Being retained may have just meant they didn't want to waste my training, or that I'd had less time to fuck up already, but I it may also have meant I was more persistent or could take rejection better or maybe was just less likely to be rejected - or wasted less time. Which is not to say I didn't have many, many reasons to question what I was doing and how I appeared.

Whatever the reason, most of what I did this point involved some real problem-solving, a little sleuthing, and a lot of dealing with doormen, supers, and building managers who only knew their non-disclosure policies and financially-motivated private ownership loyalties. Sometimes I felt that all they were trying to conceal with indignant hostility was simply ignorance. The most memorable instance was when a building super refused to even read the confidentiality agreement. When I slipped it under his door, he opened the door, crumpled it up, threw it at me, and said "make yourself a meatball". Then threatened to call the cops. We just talked to management instead.

I have some nitpicks with census humor. The big problem is that it always seems to be about nonconformity. The question is not "What is your sex?" It is "are you male or female?" I kept hoping I'd run into anyone who'd say yes to both or neither, but it was always that question we had to sort of apologize for asking. The point is, it's gender (social information) rather than sex (biological information). Although apparently the census really makes a point of reminding people that their babies are people. Especially when it's not their baby.

There's always that question that's more of just a thing we had to remember to fill out than something we always asked: was there anyone else who may have been staying with you on April 1? And a list of options including foster children, relatives, and transients "without a permanent place to stay".

I think my favorite part of training was when someone started making assumptions within an example, and our instructor said, "you're making sense. Stop it." Fighting the superstitious qualities of assumption is one of my favorite issues: stop making sense stop making sense. It actually was a bad idea to fill in gaps, because there really is a wide variety of living arrangements. In NYC especially there are a lot of the very fun-to-say WHUHE (woo-hee) situations: Whole Household Usual Home Elsewhere. Families keep their Queens toehold and have relatives staying there while they live in another country entirely, or travel for work, etc. In practice, if we made too much sense, and it was wrong, we actually would get less information, sometimes because it alienated our enumeratee. But again, we had to apply sense as to when to make sense. Sense is sensual, logic embodied.

I don't know whether it was because the objective of the census triggered my pre-existing scientific fundamentalism, or I was just really happy to have a job that required brains, but I got annoyingly passionate about justifying counting people. A friend admitted that he lived in a non-residential building and would likely not be counted, and it drove me nuts, especially when it appeared that, like many hip kids these days, he'd rather not officially exist. Often in these cases a reluctance to reveal the illicitness of one's living conditions is cited as the justification for reluctance, and I've become severely frustrated that I lack the rhetoric to explain the isolation of the census from any authority who would care.

I know that my emotional and intellectual involvement in my tiny uninformed corner of the information-constructor makes me exceptional, and is the attitude most likely to prevent exactly the desultory functioning for which I advocate. The high-functioning stupidity required is a role all of us had to play, and I suspect it doesn't matter how thoroughly one obervational tool observes when it can't observe another part of the machine. If I used body metaphors here I think that would really defeat the purpose - the more mixed the metaphor the better, when we know that what we're talking about is a complicated social interaction of humans, paper, and computers (and a few others).

The enumerator handbook writers certainly don't know how to put it. Their dialogue of escalating refusal ends with a history lesson. Of course, saying the U.S. census started in 1790 only cements its status as symptom of the modern state. My plan was to appeal to peoples' natural curiosity, but it turns out curiosity is anything but natural or even common. "How do you imagine we know how many people exist at all?" I want to ask. Democracy itself is a minority, and people just don't see themselves as makers of the information they need. I grew up with Sim City, in which the headcount was just a computer calculation. My mind wants there to be a stable truth, not one generated by and even contingent upon the very cultural attitudes and fluctuations it records.

"When you read a statistic, do you consider yourself to be included in it?" I want to ask. Are those who don't think anyone needs to know how old they are or where they live the same people who lament that statistics they hear don't seem to represent their everyday reality, or seem too fuzzy? The census advertising focused on "the community", which automatically doesn't make people recognize how defined they are by their geographical location, but implies something they already feel separate from.

It should be less difficult to make the connection between micro and macro, but the way that connection is made - pulled through the bureaucratic medium toward the center of the katamari ball - stymies our social minds because not even the starseed can see the big picture. It is always hard to understand how a series of individual decisions add up to government. Conspiracy theorists feed on this kind of psychological gap, because this is how evil is done without anyone to blame - and how individual lives are improved without any particular hero. I do wonder if the moral reliability of census workers plays a significant part in how well the count improves local services. The causal relationships are extremely shakey, and it's best to keep causality out of things like this. I actually feel that the system works based on indifference to the values of its cogs. It is, in fact, the indifference of the cogs to what they are doing that contains the minimal information, keeping them from either getting it from anywhere else or giving it to anyone but their immediate superiors. The most critical part is that the information actually isn't useful except on the wider "general statistics" scale for which it is intended.

Janet Bruesselbach, July 2010