Legible and illegible tasks in organisations

I’ve been debating with my tech lead about the role of qualitative methods, continuous discovery, interviews, observational studies, in generating user insight. In his view: if we already collect vast amounts of data, isn’t it more efficient to analyse what we have rather than spend time talking to users? Quantitative data is more measurable, comparable, and easier to use as evidence. After all, what can one interview reveal that hundreds of tagged events, funnel charts, or aggregated survey scores cannot? An interview in contrast is like a needle-in-the-haystack, does it even matter when huge amounts of data is already available?

To me clearly, this was not true. In a nutshell, the reply I provided was that both “legible” and “illegible” work is important. To explain what this means, and how it applies, I need to take a long-winded detour into the world of James C. Scott, and help you see this through his eyes.

The key idea from his book ‘Seeing like a state’ translated to organisations is as follows: As companies scale, they prioritise legible work over illegible work. This makes it harder to incorporate “illegible” work through product, processes and people. Failing to incorporate the illegible leads to a lower efficiency and throughput for the organisations.

If we were to translate this into three principles:

1. Modern organizations exert control by maximising “legibility”: by altering the system so that all parts of it can be measured, reported on, and so on.
2. However, these organizations are dependent on a huge amount of “illegible” work: work that cannot be tracked or planned for, but is nonetheless essential.
3. Increasing legibility thus often actually lowers efficiency - but the other benefits are high enough that organizations are typically willing to do so regardless.

The other way to put this is that “what gets measured, gets managed”, and when this company prioritises the measurable (more manageable) work, the overall efficiency of the company gets affected.

I’ll highlight some examples from the book to illustrate these points:

• Turning a forest into a simple grid of identical trees for easier measurement: government replaced complex topologies of natural forests into perfectly measureable, easily taxeable outputs, that are more straightforward when it comes to “calculating the inventory”. But it destroys the efficiency of the forests as the (soil fungi, insects, microclimates etc) are invisible to the planners, eventually making the overall forest sickly, and low-yield.
• The USSR experiments on agriculture: under the centralised Soviet state, agricultural produce was made fully “legible”, everything could be seen, counted, directed, and reported upon. Efficiency again collapsed as paper work mattered more than crop quality, incentives encouraged false data, and not good harvests.
• In the 19th/20th centuries, we had Scientific Management under ‘Taylorism’ that led to reducing skilled labour into measurable, timed micro-tasks. Factory workers who had a series of jobs where measured based on how much time each job takes. The modern day counterpart for Taylorist factory management would be the excel timesheets, quite common in consultancy companies where projects are commissioned based on the “time alloted” by the team. For the sake of legibility, output was prioritised over outcome.
• Hunter gatherers to early-agrarian societies: anthropological evidence showed that early hunter gatherers were much more well-built, less disease-prone and nourished, compared to the early-agrarian societies. What would have been the case? I was of the impression that progress to agriculture might have lead to more dietary diversity, not the other way round. The answer here also lies in James C. Scott’s key idea around legibility and illegibility. Early farmers could only cultivate 2-4 crops at scale especially carbohydrates such as wheat, barley, rice and potatoes. This made food measurable and controllable for the farmers, or the union of farmers, but dietary diversity collapsed. Again, being “easy to tax”, “easy to store”, “easy to collect”, “easy to standardise” and “easy to distribute” took preference. Being legible came at a biological cost which was ignored.
• This applies everywhere, even the introduction of standardised units such as the S.I unit system for measurement, led to a loss in efficiency. Previously, we had local, contextual units that were applicable only to certain regions. For example, we had this unit called the “Morgen” that was defined as the area which one farmer + one ox can plough in a morning. This was replaced by “0.086 hectares”, what do you think is more efficient and applicable for the farmers of that day?

In the world of corporate-speak, organisations and softwares, this is made so visible in terms of program status update calls, bi-weekly delivery syncs, monthly OKRs, CSATs, NPS scores etc. A company that is known for certain “illegible” qualities as a startup, as it grows to be a scale-up sometimes loses them for the sake of becoming more measurable, losing itself and it’s DNA in due course. But once in a while, you might also see exceptions to this rule where organisations consciously make the illegible work more visible:

In my previous product role at Noora Health, a healthcare non-profit, we were running a program that trained patients and caregivers in maternal child care. In this case, the number of “trained patients” was legible work. This was the figure given to the grants, philanthropists and other stakeholders who were financing the program in proportion to the scale of such impact. But you also had “number of trained patients where nurse empathetically listened to the patient, and spoke to them gently” which was very illegible to measure and manage. As a consequence, this didn’t get measured, and therefore, not managed. How can you even measure “care”? But that doesn’t discount the fact that illegible work is not important (it’s important but it’s difficult to measure).. 

To ensure that the “illegible” work was still incorporated into the core program while it continued to scale and serve millions of patients and caregivers, the caregiving program had two components: an in-person component where nurses were trained to educate patients and caregivers in health facilities. And a digital component which provided just-in-time information through mobile phones for family caregiving.

If it was all about maximising “number of patients trained”, Noora Health could have very well just revamped the program focussing on JUST the digital component (as it’s easy to scale to 10s of millions this way, why bother to have an in-person component? even..). But they didn’t make that compromise, as there is a lot of “illegible work” when it comes to providing care. The way they have absorbed “illegible work” as a part of their core-program is a beautiful lesson on how it’s possible to also scale while ensuring the illegible work remains uncompromised.

Coming back to the original premise of this essay, where I was debating with my tech lead on why user interviews are important despite tons of quant data… it boils down to the same principle: illegible work should remain uncompromised.

Data can tell you who clicked on what, what they abandoned, how often an error happened etc.. but not “what users wish they could do”, “what they tried but didn’t know how”, “what workarounds were created” etc.. In that sense, data is a lagging indicator. It measures the past, and not what’s possible. What’s possible is not captured in “legible” ways. We have tacit knowledge. Lived experience. Informal behaviors. All these are “illegible”.

Data only shows the legible past. Continuous discovery makes the “illegible” work more visible, even before it becomes data. It’s not to say that this is a good replacement for quants. It merely completes it, it’s another lens we can adopt for truth-seeking, building products closer to what users really want.