2028 will mark ten years since Richard Thaler coined the term sludge. While not quite here, the thought of this anniversary being so close does seem surprising – behavioural sludge still feels like a new idea. Certainly, it is an idea that is gaining momentum within behavioural science, attention from practitioners, and interest from the public at-large. There is still much to be done though. In this blog, Stuart Mills highlights areas where sludge research could and should go next.
What is sludge?
Sludge seems to have emerged as the antithesis of nudging. Where a nudge helps people achieve their objectives (as judged by themselves) by making things easier, sludge hinders people in achieving their objectives by making things harder. Common examples include excessive paperwork when applying for a government provision, or the use of voter ID at the ballot box – particularly when voter fraud is miniscule. Hence, Thaler encourages us to make things easier, not harder. To nudge, not sludge.
However, sometimes we might be more likely to achieve our objectives when things are harder. A fun example comes from research by Dilip Soman and colleagues around ‘decision points.’ They discuss how smaller popcorn boxes could encourage people to eat less popcorn, because refilling the smaller box requires more trips to the concessions stand. For someone on a diet, this friction may actually be a good thing.
There are more examples. Both borrowers and banks probably benefit from the hoops one must jump through to get a mortgage. Gambling is another important area to consider – making it harder to start gambling, but easier to stop, is probably sensible. Or, consider education, where research shows the important role friction plays in learning. To be sure, students will feel smart if we just give them the answers, but they’ll feel a lot less smart when they fail next week’s test. Friction may even be necessary to fully appreciate art!
To be sure, it’s a bit of a straw man to say that sludge scholars believe all friction is bad. For Thaler and Sunstein, friction only becomes sludge when it gets in the way of people doing what they want to do. For the person who really wants popcorn, the walk to the concessions stand is sludge. For the dieter, it is a handy nudge, analogous to the famous moving of snack food away from supermarket check-outs.
Most work on sludge, then, implicitly holds it to be subjective. Yet there also seems to be something objective about it as well. We all experience sludge, and sludge can evoke fierce emotions when encountered. Excellent work linking sludge to transaction cost economics has done much to enhance ideas around what sludge is, or perhaps more appropriately, what should and should not be considered sludge. But this tension between a very subjective definition, on the one hand, and a feeling of something more objective and tangible, on the other, is an important problem to understand.
Behavioural audits and de-sludging
One reason why this tension is important is its relevance to the emerging work around behavioural auditing and de-sludging. There has been a call for ‘sludge audits’ for almost as long as the concept of sludge has been around. This auditing literature makes two implicit assumptions about sludge. Firstly, that sludge is measurable, in some form. Secondly, that once identified, sludge can be removed or ameliorated.
Sludge auditing remains an emerging area of development, but much has already been achieved. There have been audits into gambling websites, access to healthcare, and online commerce. Audits have explored different methodologies, from standardised checklists to mystery shopper approaches to mapping a person’s ‘behavioural journey’ as they navigate different websites. New technologies, like artificial intelligence, promise to scale auditing approaches quite significantly.
Figure 1: Audited Behavioural Journey for a Spotify user
Source: Mills et al (2026)
Nevertheless, the subjective nature of sludge lingers in the background, and this has important implications for the auditing programme. Specifically, who do auditors have in mind when they undertake an audit? Websites that make it difficult to cancel subscriptions might be sludgy, but they’ll probably feel less sludgy for someone with high digital literacy. Having to pay many little service charges, rather than one cumulative charge, is also sludgy, but it probably feels less sludgy for someone with ample disposable income. And having to fill in multiple forms, perhaps disclosing the same information several times, is (again) sludgy. But less so for someone who can pay a secretary to do it, and more so for someone who is caring for two toddlers.
For all the progress the behavioural auditing literature has made concerning sludge, we are at risk of falling into the infamous Harvard psychologist trap – we know an awful lot about the sludge that behavioural scientists experience!
What should we be asking?
This ‘who is the user?’ question is an important one for sludge auditors to answer. Here are two more.
Firstly, if sludge is a subjective feeling of burden experienced from onerous systems and processes, it may be less helpful to think about a quantum of sludge (e.g., an additional form) and more helpful to think about a continuum of sludge (e.g., the process of form-filling, as a whole). Having to complete form A or form B might be sludgy. Having to complete form A and form B is likely to be more sludgy. But there is no reason to think the total ‘sludge’ would be A+B. Indeed, B might feel even more onerous because a person had already had to complete A; there is no reason to assume that our patience is exhausted at a constant rate. This perspective is what I have previously called ‘cumulative sludge. At present, we know very little about how the order of sludgy choice architecture impacts the overall feelings of a process.
Secondly, sludge auditors should give more attention to de-sludging. Having undertaken sludge audits, it is common to complete the audit and then be faced with the terrible question: so what? Much of the literature on sludge auditing has (probably rightly) focused on developing methodologies, but too little has been said about tackling sludge once it has been identified. In some instances, behavioural science interventions are probably worthwhile. Sunstein, for instance, has argued that better use of default options might eliminate a lot of paperwork in government processes.
But there is good reason to think that nudge is not always the antidote to sludge. If I commit a crime, I must deal with the complexities of the law. The law is sludgy. Ergo, we might choose to make it easy by simplifying the language and giving me some tips on how to act in the courtroom. These sludge-busting techniques would make it easier for me to represent myself. But I’d still go to prison! Governments around the world recognise that, because the law is complex, and often necessarily so, the best solution is to provide defendants with legal representation. How to intervene, once sludge has been identified, is an essential question for sludge scholars to tackle. Behavioural science interventions may not always be the best response to sludge.
Towards a behavioural science of friction
The idea of sludge has also revealed the importance of friction in human behaviour. Indeed, there is the potential for friction to become the main character of behavioural science in the years ahead.
As above, discussions of sludge have prompted various debates about when adding friction to processes can improve outcomes. Yet it is also evident that too much friction can lead to poor outcomes, be it people abandoning whatever it is they want to do, or achieving their objective, but being stressed out and irritable (or worse) in the process. Therefore, rather than talk about nudge and sludge, it may be much more interesting to think about optimal friction. What might such an approach look like?
Assuming friction could be measured (I will return to this), we could plot this on the x-axis of a graph. On the y-axis, we could plot the probability that some outcome of interest, call it A, is chosen. Optimal friction would be the amount of friction coinciding with the greatest probability of A being chosen. Intuitively, if both too much friction and too little lead to poor outcomes, our hypothetical graph will show something resembling an n-shaped curve. Nudges, sludges, boosts, and so on, would all then be tools for shifting the friction of the process towards the optimal friction point.
Figure 2: Hypothetical ‘Friction Curve’
Of course, the conceptual challenge here is that unlike in physics, behavioural science lacks a standard way of measuring friction. What is the friction of an additional piece of paperwork? How does it compare to fewer pieces of paperwork, written in less accessible language? How much less friction does an online form impose, compared to having to go to the post office? What is the friction of a trip to the concessions stand?
At present, there are no immediate answers to these questions. There might never be. But the study of sludge has elevated the importance of friction as a behavioural science phenomenon. And with it, we might begin to think about simple experiments that could move sludge scholars closer to a ‘behavioural science of friction’. For instance, information disclosure forms could be systematically varied from simple to complex, measured by a Flesch-Kincaid readability score, with participants asked to make an active choice about, say, about how much of their income to save for retirement. With readability as a proxy for friction, and the percentage of participants choosing option A as A’s probability, a friction curve could be derived. It might have an n-shape. It might not. But it’s certainly worthwhile to find out.
Stuart Mills is an Assistant Professor of Economics at the University of Leeds, and a Visiting Fellow of Behavioural Science at the London School of Economics and Political Science. His research focuses on technology, behaviour, and public policy.