By Gordon Hull
At the end of my time in high school, I worked part-time bagging groceries. There was some modest union influence on the job, and its scheduling was pretty predictable: the longer you’d been there, the better schedule you’d get. Your first few weeks, you knew you’d be working late into the evening, especially on Friday and Saturday. After a while, the late shifts would taper off as somebody newer than you would get slotted into them. You could depend on a pretty predictable schedule week-in and week-out. It was a service sector job with a factory-like scheduling.
I mention this because one of the more nefarious uses of big data got emphasized last week in the context of discussion Black Friday’s steady march backwards into Thanksgiving day. Last week’s news highlighted one way that data analytics can be used to introduce further precarity into the lives of low-wage workers. The transfer of risk and precarity to employees more generally is of course something neoliberalism does pretty well, but the process is even more intense for low-wage workers due to the introduction of scheduling software that produces unsteady, uneven, just-in-time scheduling, so that employers don’t have to pay for employees who aren’t absolutely necessary. Since a disproportionate number of those affected by these programs have children to care for, and since many of them are minorities, it’s also a case of disparate impact on poor, minority women. As stores open earlier and earlier for Black Friday, more and more workers – again, mostly women – are being put into the position of not knowing whether they’ll have Thanksgiving off until a day or two before. It’s a good example of the general problem.
Here is Amanda Marcotte on Slate:
“The software, which uses store traffic patterns, weather, and other variables to produce timely estimations of how many employees a store will need during any given shift, helps save companies money, but only by forcing low-wage workers to live the "on call" lifestyle, not knowing from one week to the next when they're supposed to work. That mentality does not take a holiday for Thanksgiving, with many workers who had made travel or family plans being told at the last minute to drop those plans or get fired for not showing up to work.”
As Marcotte goes on to document, this of course causes real, quantifiable increases in the levels of stress these workers face, since it makes it nearly impossible for them to juggle their (poorly remunerated) jobs, child care and other obligations. Such workers never had it easy, of course; on a slow day at the grocery store, you could always be sent home early (and without pay for the time you were scheduled but didn’t work). But this is something considerably more intense, I think, because it furthers the processes of real subsumption, where capital extends outside the factory walls and into all aspects of life. In the old way you could say with certainty whether you were at work, or not. Capital extended into the home removes this certainty.
One of the most discussed of such extensions is the direct extension of work into the home, as in the well-worn images of dads spending their entire time on the Blackberry, even during family dinners. At some point in that process, there is a further intensification: you find yourself not just doing one job all the time, but indefinitely many jobs in-between; the job morphs into what Ian Bogost calls hyperemployment (see also here and here). So for one segment of workers, it is impossible to stop working.
Scheduling-by-analytics shows the version of this process for lower socioeconomic strata. Marx had shown how capital depends on creating a “surplus population,” that then could serve as an industrial reserve army of contingent labor. Those workers would be called into the factory when there was extra work to do, and left unemployed and near starvation otherwise. Here we see the transformation of the industrial reserve army into something fitting the needs of post-industrial, service sector capital, abetted by analytics. Big data is very good at segmenting and regrouping formerly opaque-looking blocks of things – time, populations, etc. – and here we see it being used to precisely that effect, segmenting and reconfiguring the time of the working day to align that time as precisely as possible with the needs of employers.
In the factory system described by Marx, it is the steam engine that dictates time. In the contemporary service sector, it is predictions about customer traffic, and producing very granular predictions is the service the new scheduling software provides. The result is both an extension of work into the home that is almost the mirror image of the Blackberry dad, and also an intensification of the production of surplus population. The old Marxist surplus population had an excess of time away from the factory; the beleaguered service-sector employee can’t escape into her house or otherwise be away from work, not because she is working, but because she is not. In other words, the low-level service sector worker cannot escape work, even if she isn’t actually working, and even if she won’t actually be called into work, because the boss might decide at the last second that her services are required (temp workers have had to put up with this for a while, of course; the disturbing part here is these are workers who have “regular” jobs).
So: for one tier of the post-industrial labor force, hyperemployment. For another - the lower classes - hyper-reserve.