The first computer I ever used was an 80286 with a 5 megabyte hard disk and an EGA card. It ran on MS-DOS 4.0. Today, I use a two-generation-old iPhone 4S. It has orders of magnitude more power and connectivity than the first computer I ever used and it’s smaller, but what makes it truly revolutionary is the way it fits into my life. Compared to my phone, my first computer – indeed, even my current personal computer – is a pool of abstract computing resource that I have to turn on or off, running an operating system I had to buy and install. I sat down to use my computer like some kind of occasion or event – I still do.
My phone, in contrast, is a seamless background part of my life; it’s an extension of my will to communicate and perceive the world, because it’s how I send and receive texts, Facebook posts, phone calls, emails and it’s always on, always next to me. Despite the vast differences between how these two machines fit into my life, however, we have only one real word to describe what I am to it – a ‘user’.
The most significant advancement between my 80286 in 1982 and my iPhone 4S in 2014 isn’t the level of miniaturization and power in my iPhone; rather, it’s the way that it fits into my life and lets me access and interact with the big data backend that makes my phone run. For lack of a better way of putting it, we’ve termed this user experience – an all-encompassing term describing how technology becomes an adaptive part of people’s lives. But this is problematic at best: since my phone fits into how I work and play and in general get on with life, it’s obviously a fundamentally different experience for me as a ‘user’ than my computer was (or for that matter even is today, since I still use a desktop computer).
User Experience Is Increasingly A Bullshit Term
Driven by the twin engines of intellectual fashion and cottage-level careerism, “user experience” today is a trendy term that is freighted with old assumptions that no one wants to examine. Chief amongst its easy and unquestioned assumptions is the assumption that every human that interacts with a machine is its ‘user’.
If you really think about it, it is almost laughably easy to shoot irreparable holes in this terminology. What if humans are the raw material that machines work on? This covers machine systems from tattoo machines to hospitals to military recruitment systems, employee recruitment, people-tracking and education; the people in these systems aren’t really ‘users’ so much as raw material inputs into a system producing finished trained people as its material output. What about systems that essentially custodians for humans – hospital bed check-in and check-out tracking, blood pressure and biometric monitoring in emergency rooms, insurance regulations? Hospital patients and insured customers cannot be thought of as the primary ‘users’ of these systems without stretching the term ‘user’ to encompass more than it was originally meant to, since most of them don’t even touch the machine’s controls or alter its function – and what about the medical staff and nurses and doctors, aren’t they also ‘users’? Does such a system even care about the ‘experience’ of patients in such as setting, or are they simply more ‘users’ along with the doctors? What about machine accession systems at museums — are the artists the primary users or are the patrons, or the museum staff? Which user does the data ‘belong’ to in these circumstances, the system administrator? What if there is no system administrator?
‘User’ is heavily freighted with connotations that are not always appropriate. Originating from the earliest multi-user Unix mainframe installations in the 20th century, it assumes some kind of permanent record with a system (a ‘user account’), a permissions status (read/write/delete view, expressible as a hash like in Linux), a user group, and a system administrator. None of these apply to my phone: I’m the only user so the only permanent record record on the phone is mine, the phone is literally the account itself. There is no permissions tracking on my phone – though it prompts me for passwords and security information occasionally, these are mostly for services my phone delivers me, not my phone itself. There is no system administrator on my phone – again, I’m the only ‘user’. I don’t have a walled-off ‘user area’ on a shared machine because I’m the only user on my machine; I have no ‘user files’ since all the file are mine; and a vast majority of the time, my ‘experience’ with the phone isn’t even direct ‘use’ – it’s just sitting there on my desk and chiming occasionally.
It’s not just that my phone and the world of data that I interact with through it are more advanced – the relationship I have with the phone and that data are also markedly, substantially different than the way I interact with a computer.
In big data, ‘user experience’ is an even bigger canard than in software at large. The entire point of machine data is that it is not predicated upon consumption by a human. Whether you are there or not or watching or not, it still goes on and it still has value. Machine data isn’t some kind of Schrodinger’s cat in some state of quantum flux when you are not observing it, because it’s an orderly process that you’re observing yourself directly or indirectly in the form of programmed rules. It’s not about you or your experience; it doesn’t even care if you’re there or not.
The philosophical task at hand in big data is to resolve the systemic mismatch between the abstract and historically recent notion of ‘user experience’ and the modern reality of machine systems consumed by groups of people who aren’t really ‘users’ — a mismatch between machines for use by business people and machines that make a business out of people. But instead, what we see in the field is that out of inertia, commercial interest in trademarked terms or simple commitment to language without really thinking about what that language means (a hideously common mistake in companies run by technical people that devalue or don’t understand language arts or cultural history), people are trying to expand the term ‘user experience’, hitching their careers to the UX bandwagon and developing lengthy university courses and coffee table books and colorful fashionable magazines on UX without even answering hard questions about the metaphysical underpinnings of the term ‘user’.
What none of these people want to hear is that the term “user experience” might be, to put it frankly if ungently, outdated, naive, careerist bullshit that doesn’t apply to big data. Sorry. A big data advocate talking about ‘user experience’ is like woodworking tool salesman talking about ‘craftsman experience’ or a lawyer talking about ‘client experience’. Selling big data by talking about user experience is like selling guns based on ‘how it feels to shoot it’. It’s nice, but it’s not really that important. The more important questions are, can you build it or not? Will it work or not? Are you going to take and win the case or not? Can you make sense of my data or not? Screw the abstract, soi-distant, mythical ideal ‘user’, let’s talk imperfect, busy, harried, stupid old me: will I get the data experience I need?
The standard answer that UX advocates have is that a good user experience design makes it easier to understand data. But this is also, upon close examination, a canard. After all, how much can the best $200 an hour UX consultant really do if you have a crappy back end? What training do people even have to understand the rationale behind UX dictates and to implement them outside the user experience design world? An awesome and friendly and exciting ‘user experience’ will not turn the stupid people in your organization into smart ones or the difficult customers you have into nice ones — you still have a service or a product you’re providing, and changing its mode of delivery is not some form of magic that makes the data driving that service or product any more meaningful or useful than it already is. No matter how you justify it, when you treat every human being that interacts with your system a ‘user’, the result is that user experience is always an aside, always an additional concern, and never the prime concern when you really question how important it is.
De-Linking Users From Data Through Data Experience: Data Intentionality
As easy and even occasionally rewarding as it might be to critique commonly held notions in a technically savvy but philosophically under-educated linguistic community, it’s beholden on anyone who makes such a critique to present some kind of alternative. (Or, in other words, even though it’s fun to troll techies groping their way around a philosophically novel technological area without the benefit of ever even bothering to take a psychology or linguistics class, sooner or later I do have to actually try to propose something useful instead of just taking you all to task for not knowing something you weren’t even supposed to know.)
To that end, I’ll propose some concepts for this area which will hopefully be as mercilessly critiqued as I’ve been to the status quo. I’ve briefly mentioned data experience, a concept I’ll expand separately. Let’s start with data intentionality.
Data is always about something — that is, it always has an intentionality, or a directedness. The standard paradigm of data intentionality assumes that it is data about the user — this is why we have notions like user directories, user permission levels and user groups. In the Unix-driven mainframe systems that shaped our language, a ‘user’ was, in a sense, a software object with properties and data pertaining to them; all data belonged to at least one user.
Today this is plainly no longer the case. Consider a common example: a team multiple data scientists might for instance study a set of data detailing the click-through and churn rates of customers using a shopping cart. For the scientists, none of the data they are working on ‘belong’ to any of them in any sense of the word ‘belong’. The data is not ‘about’ the user – the data is about the business. The scientists are also not a single ‘user’ in the sense of the word – they have different permissions and data areas within the corporate IT system they work within and they bring different analytical talents to bear. The scientists’ work is not ‘about’ their experience — whether they’re working in command-line tools and manually reading the results of
tail -f or using the latest high-priced business intelligence tools, their experience is subordinate to the result they produce — no one really cares how you did it, just that you do it.
Big data is driven by decreasing costs of commodity hardware, an increased availability in trained data scientists and a massive increase in the rate and volume of data generated by modern machine systems. The type of data in these contexts is sometimes termed “machine data” to distinguish it from data generated for consumption by human beings. Machine data makes big data businesses unlike normal businesses — rather than data being used to inform people about a business process, the ingestion, analysis and output of data as information is the business itself in the machine data context, and it’s about the business. The fundamental issue is that in ‘big data’, the data is very, very rarely intended for — ‘about’ — the user. As tautological as it sounds, the intentionality of data in big data is about the business, and the business is about, well… the data.Tags: big data, design, user experience