Emancipating Humbots
Taylorism
In her 2001 article, “Automate & Informate: The Two Faces of Intelligent Technology”, Shoshana Zuboff cautioned against the pitfalls of focussing solely on automation, and proposed a better balance between Automating and “Informate-ing” (which — roughly — describes the process of humans creating new information and product from analyzing and manipulating data). She discussed various deficiencies of automation, particularly those arising from its Taylorist roots. She also speculated on whether some of the newer (at the time) tech organizations would inherit this Taylorism or whether they would embrace informate-ing (as opposed to automating).
Frederick Taylor’s ideas of scientific management, focussed on increasing labour and capital efficiency and productivity, and hence, profits. They involved identifying the most optimal process for carrying out various tasks, and defining manufacturing flows accordingly. This led to human assembly lines where workers performed mundane, and repetitive tasks at high speed. The workers were essentially human robots (“Humbots”)
This Taylorism led to a new class of worker: the manager or “boss”. Unlike the “worker”, who had a single repetitive task, bosses had complete visibility into the factory, and what might improve its processes. They also had the authority to command and control workers, such that these processes were performed optimally. Bosses were considered “part of the organization” or “insiders” as they had ownership of processes, and could be considered loyal to the organizations aims.
With progress in technology, the human assembly lines were increasingly replaced by machines — a direct substitution of capital for labour. This aggravated an already tense relationship between workers and the organization. Workers increasingly saw their role as undignified, and temporary until the organization found some way of automating Humbot work into robot work.
Happy Employees
The open plans offices of, say, Google or Facebook, look nothing like the factory assembly lines described above. All around, one sees smiling faces pampered with free food and other perks. Unlike the workers who were commanded and controlled to do a specific set of tasks, modern tech employees operate in flat organizations where (often) no boss tells them what to do, and they are asked to pursue whatever cause that best achieves the companies “mission”. They are driven by uplifting maxims like “nothing is someone else’s problem” and “Don’t ask for permission” — implying that they can do anything. They are their own bosses to the fullest.
The freedom of tech employees goes even further. There are few restrictions to how employees can use the companies data, and as a result have the opportunity to try out new ideas, and build new products that might utilize this data. Many wildly successful products have emerged from a small number of employees taking some data and seeing “what they might build around it”. Hence, Zuboff’s idea of “Informating” is probably most vibrant and successful in the modern tech company. And similarly, no place is further from the Taylorism command and control assembly line.
However, there is another, darker side.
The Modern Humbot
I’m going to use Google as a guinea-pig, but my observations apply to many data-centered tech companies.
While the mission of Google is “to organize the world’s information and make it universally accessible and useful”, in practice, it’s core business (currently) is manufacturing. It manufactures ad impressions (which account for most of Google’s revenue).
Ad impressions are manufactured by two classes of Google employees. The first class of employees spend large amounts of time on Google properties (Google Search, YouTube etc, Shopping), performing various actions (searching, watching videos, buying things). The second class of employees build systems that make sense of these actions, and build more products that will help the first group of employees perform more actions.
Google treats the two classes of employees very differently. It is the second class that I described above as empowered, pampered, and happy.
The first class of employees, are not even officially referred to as employees. They are varyingly referred to as “community” or “people”, and in a few situations “customer”. The organization often reminds these employees on how lucky they are on getting a service for free — conveniently hiding how valuable they are to the company (not necessarily individually, but immensely in aggregate), or the darker truth that they are actually modern reincarnation of the Humbot.
The lot of the modern Humbot is terrible. It works for free. It is completely expendable, as millions of its kind are waiting to take its place. The slightest disobedience of the boss, results in dire consequences.
This is the dark paradox of many modern tech companies. They have two classes of employees: One class, which enjoys high remuneration, and intellectual freedoms, and largely operates in a post-Taylorist, enlightened informate-friendly and informate-heavy micro-organization. This is conjoined with a vast heard to Humbots which forms a Neo-Taylorist macro-organization where automating out-dominates informating.
Ending Neo-Taylorism
In “Automate & Informate: The Two Faces of Intelligent Technology”, Zuboff recommended that a Informate-friendly organization needs to 1) empower employees to take action based off data, and 2) provide them with the necessary skills, training and education to do this satisfactorily. There is a third, implicit requirement, which is that these employees should also have access to the said data.
The only way to end Neo-Taylorism in tech macro-organizations is to transform it into a more informate-friendly form. For example, what if Google Search could be transformed into its true role as an organizer of information, and all of its search indexes and people information is broadly shared)obviously with the appropriate privacy and security controls)? Google becomes, not an advertising company, but an infrastructure company, with millions of other developers building new products on its infrastructure.
There are obvious barriers to this approach. Bosses, or “Middle Managers” as Zuboff described them, have always resisted informating. In automation-heavy organizations, the boss has a privileged positioned, endowed with exclusive access to information, the support of the organization as a loyal employee, and one who has rights to command and control. The “employee” in a tech micro-organization enjoys almost identical privileges, and where “command and control” involves almost infinite flexibility on how Humbots are treated. Hence, extending informating outside the micro-organization and into the macro-organization democratizes privileges that a small elite enjoyed, and that elite will resist it.
Silver Linings
Many have proposed building informate-friendly, often “open” versions of tech giants. However, given the level of market monopoly that many tech giants enjoy, this is going to be near impossible. The failed attempts by Bing and Google+ to topple Google Search and Facebook respectively, demonstrate how hard it is, even for giants to slay giants. Hence, “building a new version of X” is probably not going to work. Others have hoped that governments will break monopolies through anti-trust regulations — though, the enormous lobbying power of giants makes these efforts usually impotent. Also, half a giant is usually still a giant — so splitting a Google into two, will hardly solve the problems expressed above.
There seems to be no obvious solution for existing products. However, there might be some hope for new products.
What if all new products are designed such that data is distributed across the macro-organization instead of a small micro organization? What if data is, by design, distributed? And control free for all the nodes in this giant graph? The obvious downside of this approach is that no single entity will have exclusive access to data, and hence will not become a giant like today’s modern tech giants. On the other hand, this system would be far more beneficial to the majority.
While today’s modern tech-giants are the fruits of capitalism, they are themselves highly anti-capitalist. They stifle free markets and thrive on monopolies and anti-competition. They hoard and hinder the movement of the most important form of capital: data. An informate-friendly distributed market is the ultimate free-market of data. It is the ultimate form of capitalism. And surely, the best future for humanity.
