This Field is Required
Tracing the origins of 'life admin' from sacrificial servants, to human computers, chatbots and Potemkin AI.
Originally, the word ‘computer’ referred to a job title. It was a label given to accountants or tax officials who were ‘skilled in calendrical or chronological reckoning’. The first recorded use of the term ‘computer’ comes from a 17th century book by the English poet Richard Brathwaite who’s personal computer, the ‘best Arithmetician that ever breathed’, had ran the numbers and calculated that the span of a human life was ‘threescore and ten’ (70 years). As it turns out, Brathwaite’s computer had cribbed that number from the book of Psalms – the Bible being the Stack Exchange of its day – but the point he was trying to make was that life is short.
Way too short, in any event, to be constantly filling out forms, completing reCAPTCHAs and resetting passwords but that’s what we spend a vast chunk of our modern lives doing. If we spend roughly six months over the course of our lives waiting at red lights then how much time do we spend filling out useless forms simply because every company we have to deal with wants to run their own informal census? Don’t answer that, I don’t actually want to know.
In the 1930s and 40s early punch-card and magnetic-tape computing machines were used by the Nazis (I’m looking at you IBM) and against the Nazis (rest in peace, Alan Turing) but the occupation of ‘computing’ continued for several more decades. Viewed as ‘clerical’ in nature these roles were often performed by women because, at least in ‘The West’, female admin staff could legally be paid less than their male peers. In the last few years some of the human computers that worked on NASA’s Apollo program have received some belated recognition courtesy of Margot Lee Shetterly’s book Hidden Figures (also adapted into a somewhat saccharine film in 2016). Meanwhile, in the Soviet Union, a greater emphasis on gender equality meant that the human computers gradually transitioned into digital computer-programmers – neatly inverting the male-dominated programming demographics of Silicon Valley.
With each breakthrough in computer science the same question has been asked – do we still need people? Ever since Asimov conceived of the Positronic Brain the promise of ‘real’ artificial intelligence has been spruiked as the next imminent technological revolution while the actual technology has always remained just over the horizon.
Caught up in the hype and subject to continual pressure to appear ‘cutting edge’ corporations have responded by faking AI until someone else makes it. Typically this has meant employing hundreds of poorly paid and overworked data-entry and admin staff to stand behind the curtain like the Wizard of Oz and make it seem like their expensive machines run all on their own.
This is especially true of large corporations like banks, insurance agencies, utilities and telecom providers. Anyone that deals in hundreds of thousands of customers is usually freighted with legacy ‘back-end’ systems (databases and lists of policies/accounts etc) that are well past their use-by date but extremely expensive to upgrade (a survey by Harvard Business Review put that the average cost of major IT projects like this at $167 million USD). This process of moving from an old database to a newer one is referred to as ‘data migration’ but, unlike in the natural world, where it’s understood that a certain percentage of critters will inevitably get lost or eaten along the way, losing customer records in the business world is strictly verboten – often incurring a whole host of liabilities and resulting in what corporate lawyers refer to as ‘reputational damage’. So the old systems remain – buried under more and more layers of interfaces, caches and ‘data services’ that all rely, at heart, on that same glorified rolodex.
The point of mentioning all this is to provide some explanation for why interacting with large organisations on the internet is so consistently frustrating. It also explains why we’ve ended up doing a great deal of work on behalf of these dysfunctional organisations. What’s now known as ‘life admin’ has come to be everyone’s second (or third or fourth) job. Although it’s difficult to get a sense of how much time is sunk into registering for, and managing, various online accounts, in the last six months, the browser on my laptop has helpfully recorded a whopping 123 sets of credentials for everything from myGov to MIFF. As the late, great David Graeber stated in the opening to his book The Utopia of Rules:
“…we live in a deeply bureaucratic society. If we do not notice it, it is largely because bureaucratic practises and requirements have become so all-pervasive that we can barely see them – or worse, cannot imagine doing things any other way”
Forms, Forms And More Forms
As a society we are drowning in forms. We fill out forms to lodge our taxes, rent homes, hook up utilities and book appointments but we’re also forced to fill out forms to use websites, watch TV, buy things on the internet and activate the things we’ve bought in real life – after which we’re expected to fill out forms to rate our satisfaction with those previous forms. COVID has pushed us even further toward some sort of paperwork singularity – for a brief period between the outbreak of the pandemic and the adoption of QR codes, we had to fill out a form just to visit a cafe.
Again Graeber reminds us that the promise of the internet, and of digital technology more generally, was to liberate us from menial work by automating it away. Instead the transition from actual ‘files’ and ‘folders’ to their digital equivalents appears to have drastically increased the number of bureaucracies.
“Just as the invention of new forms of industrial automation in the eighteenth and nineteenth centuries had the paradoxical effect of turning more and more of the world’s population into full-time industrial workers, so has all the software designed to save us from administrative responsibilities in recent decades ultimately turned us all into part-time or or full-time administrators.”
These days you can gauge someone’s social status by measuring the amount of time they spend filling out forms. At one end of the spectrum are the people that rely on paperwork simply to survive. Those dependent on ‘means tested’ welfare systems are undoubtedly the most acute victims of bureaucratisation (part of what social theorists refer to as ‘structural violence’). In Australia the ‘social security’ system is run by US firm ‘Concentrix’ (an offshoot of IBM) under the brand ‘Centrelink’. For those forced to interact with Centrelink the experience can vary from mildly frustrating to deeply traumatic but this dysfunction should always be understood as a feature, not a bug in the system. As its name suggests, Concentrix specialises in making babushka dolls out of red tape.
This is well understood by those forced to interact with Centrelink even if it remains a poorly kept secret amongst the wider public. For the sake of propriety most mainstream commentary pretends that the system simply suffers from occasional whoopsies. A web search for ’Centrelink’ reveals credulous headlines like ‘How Centrelink’s ‘robodebt’ ran off the rails’ and ‘Centrelink recipients caught in another bureaucratic bungle’ as if the point of these systems wasn’t to discourage people from applying for welfare.
The current Services Australia website lists 275 individual forms – each more obtuse than the last. Examples include the SA381 form ‘Carer Allowance Questionnaire – Carer not living with the person for whom care is being provided’ or RS017 ‘Random Sample Surveys Partner details form’. Others, like the ‘Request for restoration of Disability Support Pension or Age Pension Claim for Crisis Payment and Anticipated Payment’, seem designed to trick the elderly into thinking they’ve had a stroke.
Centrelink forms are clearly designed to be punitive but job-applications forms are rarely an improvement – often requiring you to dismantle your C.V. so that the info can be wedged into a ‘machine-readable’ format. Meanwhile us humans are constantly obliged to complete reCAPTCHA tasks so that the next generation of Google self-driving cars can recognise stoplights and give-way signs. Large companies are also notorious for demanding applicants complete lengthy psychometric personality tests – forcing us to indulge in the Astrology of the white-collar workplace.
At the other end of the spectrum you have the ultra-wealthy and the upper echelons of the professional managerial class (together with celebrity actors and sportspeople) who have almost all their day-to-day affairs managed by a retinue of assistants, accountants, managers and lawyers. We get an occasional glimpse of what it’s like to be in this rarefied position whenever one of these figures discovers that they’ve been fleeced by their manager or financial advisor (see Nicholas Cage, Elton John et al.). Likewise this same separation from any sort of budgeting or life admin leaves politicians vulnerable to ‘gotcha’ questions about the price of everyday groceries – inevitably triggering a deluge of Lucille Bluth memes. Still, the disadvantages of being exempt from all of life’s menial clerical tasks are clearly outweighed by the advantages. Ghengis Khan managed to administer the largest empire the world has ever seen without learning to read or write. Eight hundred years later Donald Trump proved you can do the same thing without even being able to fully articulate your thoughts.
For the rest of us, life admin generally consists of navigating a growing number of online ‘self-service’ portals which are usually pitched to us as a convenience. ‘Did you know you can manage your account online?’ the chirpy, pre-recorded voice reminds us over the phone as we patiently wait for the opportunity to un-fuck our life in some small way. Within the corporations themselves these self-service systems are part of a general trend towards ‘automation’. But when companies talk about automation they’re talking about their processes, not yours. For a large utilities company or a telco provider tasks are considered ‘automated’ when they don’t involve human input at their end. They haven’t removed the burden of administration. In most cases they’ve simply transferred it to you (see also: the now ubiquitous supermarket self-checkout). This is why life can feel like an endless series of attempts to remember what password goes with what website.
Take the process of moving house as an example. When you need to connect to electricity and gas at a new address the easiest thing to do would be to contact your utilities company and tell them where you’re going and when you need to be connected. But experience has told you that you’ll be waiting on hold until the sands of time have scoured the great pyramids of Giza to the barest nubbins. So, reluctantly, you go to their website to do it yourself. There you’ll have to identify who you are, prove that you’re not a robot, register or remember your account credentials and then fill out a long form that probably won’t allow for an overlap between your move dates and struggles to verify your new address. The form won’t adequately explain why you can’t stay on the same plan, it won’t factor in the time you’ve spent as a customer and won’t allow you to consolidate the last bill from your old address with the first bill at your new address. You’ll probably have to re-enter your credit card info and the form won’t allow you to transfer the old plan to your housemates – only disconnect the poor bastards. This is because, from the utility company’s perspective, you’re not a person moving house. Instead you represent a meter number that has to be decoupled from a billing system while, coincidently, another meter number has to be paired with the billing system. The fact that the same person is involved doesn’t really have any bearing on the process.
This structural stupidity is all the more galling because, in Australia at least, utilities companies only have one job – account administration. For the most part they don’t run power stations or install solar panels. They don’t fix transmission lines and they don’t check gas meters. They just run a billing system connected to a tax haven and yet, every year, they work out ways to outsource more and more of their job to you.
Maybe The Best UX Was Right Here All Along
When given the task of improving a process like the one mentioned above designers ought to start by imagining what the absolute best possible experience could be. What would it be like if the organisation, to quote Dr Hammond in Jurassic Park, ‘spared no expense’. This is where imagination helps. Designers that have been slowly worn down by turgid corporate software projects usually think about this hypothetical situation in very limited terms. They think about features and options they could add to their form. ‘Maybe it could send a text message to confirm the new connection’. ‘Maybe we could make it available within the app’. ‘Maybe it could show a neat little map of your new house on the confirmation page’.
But if you’re after the best possible experience your first instinct as a customer was probably correct. Talking to someone over the phone would have improved the process immeasurably. The person on the other end of the line would have breezed through the verification process because your phone number gets them halfway there. They could have explained the difference in rates between different postcodes and perhaps offered some ‘back-pocket’ deal if pressed on the subject of loyalty. They would have understood the need for an overlap between connection and disconnection dates and found a way to forward you the final bill. Like you, they would also be limited by the company’s shoddy admin system and hamstrung by somewhat arbitrary legal requirements so they might not have been able to transfer the existing account to your housemate but they might have been able to get that ball rolling by scheduling an outbound call to sign them up.
In an ideal world you could imagine how talking to someone over the phone might reduce friction but most of us know from experience that phone calls just mean trading one set of frustrations for another. There’s a reason this experience is so universally unpleasant and it has to do with the way call centres are run. Almost all modern call centres are tiered according to the level of responsibility/access required to handle incoming inquiries. The first tier is usually just triage – agents at this level are only there to work out whether your problem should be escalated to someone who can do something about it. The second tier of call centre workers can actually solve some problems – or, at the very least, explain to you why they can’t. The third tier is who you end up talking to if you’re grumpy and insistent and you demand to ‘speak to the manager’. They have wider latitude to deal with complaints, thorny technical problems, unpaid bills and people who are trying to manage their ‘life admin’ through the partial veil of secrecy offered by domestic violence intervention orders. In a sense this third tier is also performing a sort of triage on behalf of the company – attempting to short-circuit disputes before they enter the realm of regulators and ombudsmen or the courts.
But it’s those tier one agents that bear the brunt of our frustration at shitty corporate policies and clunky admin systems. They are the human buffer between corporations and their customers and all they’re given to defend themselves are a few basic scripts but no way to actually resolve any issues. This inevitably means that we’re forced to explain our problems to someone who can’t do anything about them only to be put on hold and stuck in the queue for tier two. To make matters worse, many companies employ agents from India and Bangladesh to handle these initial inquiries. Thus South-Asian accents and South-Asian people come to be associated with all the hassles and annoyance of corporate cost-cutting. In the last fifteen years short-staffed call centres designed to push us all towards faulty online self-service portals have done more damage to race relations than Mein Kampf.
Ultimately all digital self-service projects boil down to this same essential conclusion- the best customer service is provided by another human being through the medium of conversation. The ideal, spared-no-expense, ‘user experience’ is one where someone takes your request and quietly and unobtrusively does all the work for you. This is why corporate executives have EAs and it’s why Flight Centre inexplicably survived the advent of online bookings. Twas ever thus. The early Pharaohs clearly suspected that death was no escape from daily chores so they had their servants executed and buried alongside them when they died. Happily, later Pharaohs appear to have been buried with effigies instead of their actual staff – evidence, surely, of an early win for the trade union movement.
A personal concierge service was pretty close to what everyone relied on in the era before the internet. Customers at the first grocery stores used to hand over a list and wait for the clerk to come back with all the items. A hundred years later we’re not only collecting everything on our own but we’re also personally running the checkout process. Soon we’ll have to lodge periodic Business Activity Statements on behalf of Woolworths to lighten the load on their accounting department. Unfortunately the most recent generation to start paying bills is the first generation that hasn’t personally experienced genuine customer service. These people, as Graeber warned us, won’t be able to imagine doing things any other way.
Which is a real worry because, over the last decade, more and more aspects of our lives have come to rely on some sort of interface with the web. That means we’re now in a position where shitty web design can have real impact on our wellbeing and the assumption that there is always a human fall-back option is becoming less and less certain. More recently, as many corporations have prematurely declared ‘mission accomplished’ on their automation programs, they’ve also gone to great lengths to hide actual phone numbers.
I’m Sorry, I Don’t Understand Your Message
Online chat would appear to represent a nice compromise between phone-based customer service and online self-service. Using the little chat prompts stuck to the bottom of many websites you can, theoretically at least, ask a question in your own time and get a response from a human being without waiting on hold. They can direct you to certain pages and you can read their instructions at your own speed. At the very least the person you’re chatting to knows you’re looking at some part of their website so they can point you in the right direction.
But again, corporate cost-cutting rears its ugly head. The problem, from the company’s perspective, is that human beings are relatively expensive – so they want to hold off connecting you with their call-centre agents unless it’s absolutely unavoidable. They also assume most inquiries come from idiots who haven’t read their FAQ page so instead of immediately connecting you to a person they connect you to their chatbot. Of course it’s not actually ‘their’ chatbot. These companies haven’t built this chatbot themselves. If they had those skills their self-service portal probably wouldn’t be munted in the first place. No, they buy one of the many off-the-shelf chatbot programs and simply tell it to parrot FAQ answers at you until you either give up or demand to talk to a human. These two outcomes are more or less inevitable because despite the hype around these programs, most chatbots are dumb as dog shit and can only respond to a handful of pre-set phrases and keywords.
Some, more ‘sophisticated’, chatbots have been programmed to pick up and reuse the phrases employed by the people interacting with them. These bots have facilitated some wonderful social experiments – like Microsoft’s Tay which, in the space of twenty four hours, managed to absorb enough trolling from internet edgelords to transform itself into misogynistic Nazi. Even the most clever bots can’t do much more than make rough guesses at what you want to know. Chatting with Siri is like talking to a spooky toddler that’s worked out how to enunciate – intriguing, but still years away from being genuinely helpful. As far as I can tell the best use of chatbots at the moment is making them talk to one another until one, or both, succumb to existential despair.
The reliance on chatbots to serve as a buffer between angry customers and overworked call centre agents has other pernicious side effects because, when a chatbot fails to understand your question or baulks at your fascist talking points, the system will usually hand you over to an actual call centre agent. The problem is that this transfer is not always obvious and any blurring of the line between bot and human risks turning every customer service interaction into an informal Turing Test. Sooner or later you may find yourself asking Leon from Optus to describe, in single words, only the good things that come to mind about his mother. At the end of the day there’s really no good outcome to be had from the bot + person equation. Either you end up spending additional mental energy trying to avoid offending a machine that you’ve mistaken for a moron or you end up treating an actual person like a moronic machine.
This phenomenon looks set to increase as the overhyped promise of AI bangs up against the limits of both computer science and neurobiology. Unfortunately we don’t currently know what intelligence is – which makes recreating it a bit of a challenge. In the meantime companies will increasingly resort to smoke and mirrors. Case in point – the Kiwibot. For the last couple of years students at UC Berkeley have been getting their burritos delivered by little motorised Eskies on wheels called Kiwibots. Only in this case ‘bots’ is a bit of a misnomer because, as it turns out, the vehicles are remotely-controlled by workers in Colombia making $2 USD an hour.
Absent some major technological breakthrough the future of ‘automation’ will likely consist of poor people pretending to be robots for the benefit of rich people. Witness Elon Musk – the P. T. Barnum of the 21st century – announcing that his company plans to design and build a functional humanoid robot in the next year. To accompany this announcement Musk brought out a realistic looking bi-pedal robot that, on closer examination, turned out to be a very thin man in an elastic jumpsuit. When they eventually build the real Westworld the dirty secret will be that all the android cowboys are just struggling actors with no immediate relatives.
Part of the solution to creeping bureaucracy is undoubtedly some form of universal basic income. In his book Bullshit Jobs David Graeber interviewed a ‘benefits advisor’ in the UK named Leslie who’s job involved helping the unemployed, the disabled and those on a pension navigate the elaborate systems ostensibly put in place to prevent welfare fraud. But in both the UK and Australia the incidence of fraud is incredibly small – less than 0.1% – whereas the demand for ‘means testing’ often automatically excludes an enormous proportion of eligible recipients who never apply or simply give up at some point during the process. As David Graeber pointed out in Bullshit Jobs:
“What Basic Income ultimately proposes is to detach livelihood from work. Its immediate effect would be to massively reduce the amount of bureaucracy in any country that implemented it. As Leslie’s case shows, an enormous amount of the machinery of government, and that half-government corporate NGO penumbra that surrounds it in most wealthy societies, is just there to make poor people feel bad about themselves. It’s an extraordinarily expensive moral game played to prop up a largely useless global work machine.”
Once we’ve got rid of those Centrelink forms we can start chipping away at the rest.