I apologise for every word in this post, and the capitalised 5-letter words in particular.
WORDLE is a SIMPLE word game (in US English). The aim is to identify a 5-letter word correctly in 6 guesses or fewer. Each guess has to be a real word, and you receive informative feedback each time: GREEN means you have the letter RIGHT and in the right position; yellow means the right letter in the wrong position; grey MEANS the letter does not appear in the word.
One strategy involves trial-and-error learning via 3 or 4 simple steps:
1. Use your initial knowledge of the English language to inform initial guesses, such as guessing a word with common vowels (I go for E and A) and consonants (e.g. S, T).
2. Learn from feedback on your correct and incorrect estimates.
3. Use your new information and deduction (e.g. about which combinations work when you exclude many options) to make informed guesses.
4. Do so while avoiding unhelpful heuristics, such as assuming that each letter will only appear once (or that the spelling is in UK English).
At least, that is how I play it. I get it in 3 just over half the time, and 4 or 5 in the rest. I make 2-4 ‘errors’ then succeed. In the context of the game’s rules, that is consistent success, RIGHT?
[insert crowbar GIF to try to get away with the segue]
That is the spirit of the idea of trial-and-error learning.
It is informed by previous knowledge, but also a recognition of the benefits of trying things out to generate new information, update your knowledge and skills (the definition of learning), and try again.
A positive normative account of this approach can be found in classic discussions of incrementalism and modern discussions of policymaking informed by complex systems insights:
‘To deal with uncertainty and change, encourage trial-and-error projects, or pilots, that can provide lessons, or be adopted or rejected, relatively quickly’.
Advocates of such approaches also suggest that we change how we describe them, replacing the language of policy failure with ERROR, at least when part of a process of continuous policy learning in the face of uncertainty.
At the heart of such advice are two guiding principles:
1. Recognise the limits to centralism when giving policy advice. There is no powerful centre of government, able to carry out all of its aims successfully, so do not build policy advice on that assumption.
2. Recognise the limits to our knowledge. Policymakers must make and learn from choices in the face of uncertainty, so do not kid yourself that one piece of analysis and action will do.
Much like the first two WORDLE guesses, your existing knowledge alone does not tell you how to proceed (regardless of the number of times that people repeat the slogan of ‘evidence-based policymaking’).
Political problems with trial and error
The main political problem with this approach is that many political systems – including adversarial and/or Westminster systems – are not conducive to learning from error. You may think that adapting continuously to uncertainty is crucial, but also be wary of recommending it to:
1. Politicians who will be held to account for failure. A government’s apparent failure to deliver on promises represents a resource for its opposition.
2. Organisations subject to government targets. Failure to meet strict statutory requirements is not seen as a learning experience.
More generally, your audience may face criticism whenever errors are associated with negative policy consequences (with COVID-19 policy representing a vivid, extreme example).
These limitations produce a major dilemma in policy analysis, in which you believe that you will not learn how to make good policy without trial-and-error but recognise that this approach will not be politically feasible. In many political systems, policymakers need to pretend to their audience that they know what the problem is and that they have the knowledge and power to solve it. You may not be too popular if you encourage open-minded experimentation. This limitation should not warn you against trial-and-error recommendations completely, but rather remind you to relate good-looking ideas to your policymaking context.
Please note that I missed my train stop while writing this post, despite many opportunities to learn from the other times it happened.