April 19, 2019 (2329 words)
::
And why does it matter?
Tags: big-tech
This post is day 109 of a personal challenge to write every day in 2019. See the other fragments, or sign up for my weekly newsletter.
What does it mean for a company to be a tech company?
This is something I’ve been pondering on and off for several years now, even before I took a steep turn toward being more critical of the industry (and capitalism). I used to have kind of a snobby attitude here, to be honest - if a company wasn’t actually using cool technology, it didn’t count as a real ‘tech’ company, no matter what its investors and press coverage might say. This was when I started paying more attention to all the money flooding into the so-called ‘tech’ sector, making unicorns out of any company that might convincingly pass as a tech company (with the accompanying profit margins).
One example is Hampton Creek, which produced vegan alternatives to things like mayonnaise and managed to persuade enough investors to treat it as a tech startup before collapsing in a pile of its own egg-free mayonnaise. Verdict: clearly not a tech company. Gilt, which achieved unicorn status in 2011 before being acquired for less than it raised a few years later, seemed like more of a tech company, but definitely not a pure tech company, because it was essentially a consolidated online flash sales business & it didn’t seem to have any interesting innovations. On the other hand, Amazon was primarily a retail business, but it also had AWS, so it felt like more of a tech company. Google, Facebook, obviously tech companies. Software as a service companies, also obviously tech companies. My own startup was very much a ‘tech’ company, because our product was data.
I guess I was using ‘is it a tech company?’ as a crude filter for whether I cared about the company, and so whether I wanted it to succeed. Very dumb, in hindsight. This was back when I associated ‘tech’ with ‘mostly good’ - before I thought about the labour implications of what all these tech companies were doing. I didn’t have a clue that worker organising was even a thing, back then.
Anyway, I thought about this today because I came across this great blog post by Can Duruk on this topic, which submits that ‘tech’ has become a useless moniker because of how muddled it is. He has a follow-up blog post on the advertising business model of the Internet which develops this idea further:
A key is this: human intervention is kept to a minimum. This is partly matter of technical principle (“automation is good”), a shield against ethical criticism (“algorithms are less biased than people”) but is largely driven by good old financial motives. Like most other business, personnel costs are the biggest line items in pure tech. And once you take a hit of software margins, it’s hard to give it up.
One way, then, of defining a ‘tech’ company is by looking at the human intervention necessary to keep the company running. Now, this is fraught with all sorts of issues (which humans?), but bear with me for a bit. Take a company, and ask yourself what would happen if all the people employed by the company suddenly stopped working (they went on strike, they got caught in a natural disaster at a company-wide offsite, they all spontaneously combusted, whatever). Let’s say all the physical assets of the company were unscatched. How long would the company still be able to function?
For most companies that are traditionally not considered ‘tech’, the answer is pretty obvious: they would immediately cease to function, because no one would be around to open the stores, make the coffee, ring up purchases, deliver goods, answer emails, sew the clothing, whatever. The products or services that the company provides would no longer be provided, because the human beings doing all the providing have vanished. The company would immediately cease to be a company, at least until some new CEO came in and hired replacements for all the most crucial roles.
Now consider a tech company. Let’s say, Google. Putting aside how the stock market or customers or users might react to the news of this unexpected Silicon Valley Rapture, if Google’s entire workforce disappeared tomorrow, would they still be able to carry on as a company? Well, yes, actually, for some time. Advertising auctions would continue to occur, publishers would continue to get paid, users would continue to be able to use search and maps and Gmail and YouTube. This is almost too obvious to bother saying, but the reason a company like Google can sustain such massive scale is because individual product uses are automated. No employee needs to do anything on the fly for you to get your search results.
Eventually, of course, the problems would creep in. New product rollouts wouldn’t occur; bugs would stay unfixed; site reliability would suffer as outages piled up; search result quality might stagnate because no one’s around to adjust the algorithms as the web changes. The videos on YouTube would remain unmoderated, which means they would be even worse than they are now (terrifying to imagine).
There’s probably a spectrum here, ranging from “company would cease to function immediately” to “company could function nearly indefinitely as an autonomous unit (because of technology)”. If you want to get to the really high profit margins, you want to be closer to the latter side of the spectrum. Imagine being a sysadmin who’s completely automated away the job, so you could essentially get paid to browse Reddit all day. You could even stop showing up to work and your company’s IT system would still function, because you’ve come up with automated responses to every potential contingency.
In that sense, being able to automate away what a company does is a good thing - at least, for the company. It enables the company to scale quickly. In pro-business terms, this is thought of in terms of higher productivity per worker. In Marxist terms, this is seen more neutrally as the rising organic composition of capital, because the ratio of constant capital (machinery, software, other ‘dead’ assets) rises in comparison to variable capital (i.e. human labour, which is ‘living’).
Okay, now we have a potential framework for assessing whether a company is a ‘tech’ company. So what do we do with it? What are the economic/social/political implications?
To answer this, we shall return to a point I temporarily brushed aside in the previous section. Who are the ‘employees’ who count in this thought experiment? If all Amazon employees disappeared, does this include warehouse workers or delivery workers (some of whom are subcontractors, or working for a different company that has a partnership with Amazon)? Because while Amazon’s technical stack may have been automated - Jeff Bezos is not personally recommending you more air conditioners when you buy an air conditioner - the physical aspects of Amazon’s business have not yet been. Amazon may have purchased Kiva (a robotics company) to help automate its fulfillment centres, but there are still many, many human workers involved in this process. And there are way more of them than there are, say, engineers.
What about Uber? The app doesn’t technically need the engineers to be able to function (and depending on how well the technical infrastructure is designed, it might function for quite a while without any human input), but it most certainly needs drivers. If you keep requesting an Uber but no one ever shows up, then you are not getting the service you expect from the company. But drivers are considered independent contractors, not employees, so they don’t count in productivity-per-employee metrics, and they certainly won’t get much out of Uber’s upcoming IPO (those who have been driving long enough will get a small cash bonus, meant to be used for buying stock). This will cease to be an issue if Uber ever develops fully autonomous vehicles, but that’s not going to be reality anytime soon; in the meantime, the real-life drivers who enable the company to function are sleeping in their cars and stranded without healthcare. Whether it’s fair or not, this division of labour seems pretty fundamental to Uber being a ‘tech’ company, with tech company margins.
And what about service workers on tech company campuses? Security officers, bus drivers, cafeteria workers, cleaners - anyone who keeps the office running. Their role is to support the employees who work on the product, which means that they don’t alter the outcome of the ‘mass employee disappearance’ thought experiment, but they do have an impact on per-employee metrics. Which is partly why so many of these roles are contracted out.
These cases point to the larger implications of deciding who is an employee, and whether a company is ‘tech’. Because if you took this thought experiment at face value, you could end up concluding that the full-time employees are just really good at their jobs (by automating themselves away), and so it’s justified that they are reaping most of the value (with high salaries, benefits, stock, etc). But they are not the only ones responsible for their company’s success - much of that work has been outsourced to people who, for whatever reason, don’t count as employees.
This, I think, is the crux of what it means to be a tech company with high margins. Whatever ‘innovation’ may exist typically comes along two dimensions. The first is what we normally think of as technical innovation: writing software or designing processes that allow you to increase your leverage as a worker, translating into productivity gains. This is cool, and good, though I would argue that this shouldn’t lead to increased corporate profits because it should be freely released to the world at large (but that’s a whole other issue).
The second dimension has to do with employee classification. Now, sometimes this is obviously regulatory arbitrage (and obviously exploitative), as in the case of gig economy companies like Uber/Lyft treating their drivers as independent contractors by arguing that they don’t have control over them when they clearly do (a recent California Supreme Court ruling pushed back against this practice). Sometimes this is more clever, like when Instagram was bought by Facebook for $1b despite having only 13 employees, because all the value of the social network was created by users who didn’t think of themselves as workers (this is changing - see this Atlantic story about Instagram memers unionising). Or when people volunteer to update Waze’s maps out of love for maps (Waze was bought by Google for $966m in 2013).
Going back to what makes a company a ‘tech’ company. An alternative analysis of this would be primarily cultural - as in, tech companies are the ones that are good at performing the idea of being a tech company, even if they don’t actually have the high tech-enabled productivity (and subsequently high margins) that the ‘purer’ tech companies do. An example would be Tesla, which has a greater market cap than GM and Ford put together, despite 1) being in the business of selling cars, not software and 2) having sold WAY fewer cars than the latter two (by several orders of magnitude). But Tesla is good at performing techness, so investors (and employees) assess it accordingly. On the other hand, even if techness is mostly a performative thing these days, it still has to be rooted in an economic analysis of why tech companies are worth investing in - as in, higher productivity per worker, higher margins.
Now, you could interpret this in a hagiographic way (look at these brilliant tech companies whose amazing internal structures allow employees to create so much more value!!1). But I think that elides the more important question of where these productivity gains come from, and the implications of those gains being kept within the corporation. How much of these gains comes from monetising technical innovations (which perhaps should be shared with society more broadly, rather than locked up as intellectual property), and how much comes from innovative ways to convince people to work for you without getting paid very much, if at all? Either by finding a new way to monetise something people would enjoy doing for free (and hiring engineers to make your product as addictive as possible, so they keep doing it), or by taking advantage of what 40+ years of neoliberalism have done to the economic prospects of many workers? Should we really be celebrating that?
This was another one of those fragments (alarmingly common) where I started writing without knowing where I wanted to get to in the end. I guess the takeaway is that ‘tech’ companies are the ones that are particularly good at capturing value right now, and that is why investors are flocking to put money in the sector. Capital loves ‘techness’ - it loves being able to monetise new innovations, especially when those innovations enable greater surplus value extraction from workers. A few workers benefit from it, too, because the price of their continued acquiescence to the system (a good job with benefits and stock grants) is low compared to capital’s potential upside.
My suspicion - and really, the thing that keeps me going, that keeps me writing day after day about the problems with the tech industry - is that the world would be better off if tech companies were not able to capture all this value. You could imagine different governance structures, or funding mechanisms, or intellectual property regimes, that would enable the benefits of technology development to be available to all, rather than monopolised within the company that happened to employ the people who worked on it. And you could imagine workers having more collective power so that tech companies wouldn’t be able to get away with underpaying those whose skills it sees as disposable.
I am aware that these proposals, together, imply undoing basically the entirely telos of the modern corporation. I’m okay with that.