Evidence-based politics

Introduction

In other writings, I have repeatedly used the phrase “evidence-based politics” to refer to an approach or mindset that is virtually absent in current politics but would be of immense value.

The phrase, a semi-joking parallel to “evidence-based medicine”, is both vague and informal. (And not necessarily the optimal choice for the long term.) Here I will elaborate on what I actually mean.

For now, this page is far from a full treatment: my own thoughts of the details have not necessarily progressed far beyond a vague ideal and a full eventual treatment would likely be far longer—potentially, book-length. (But, if so, this book will not be written by me.)

I caution the reader not to draw too much on “evidence-based medicine” when trying to understand “evidence-based politics”.

The problem

Politics is filled with poorly conceived solutions and “solutions” for this-and-that. In very many cases, it is clear in advance that the poorly conceived actually is poorly conceived—provided that we draw on scientific knowledge, history, economics, experiences from earlier attempts to do the same, or whatever might be relevant in the given situation. In many cases, the politicians have been outright warned that something would not work, would have too negative side-effects, or otherwise would be unwise. The politicians have still pushed ahead.

It might even be that the original problem to be solved was no such thing, e.g. because it was something that market forces would have handled well without government intervention, that the citizens could have done for themselves without a nanny government doing it for them, that would have blown over by it self, or that was too small to bother with—while, respectively and e.g., the government intervention sabotaged market forces, the nanny government did a worse and more expensive job than the individual citizen, the side-effects of the “solution” lasted longer than the short-lived “problem”, and the cure was worse than the illness.

It is important, however, not to confuse a mistake in implementation with an attempt to implement something that someone happens to dislike: Different individuals, be they politicians, voters, or something else yet, have different preferences and priorities. Not everyone will agree on what a reasonable goal is, and the focus here is on poor attempts to reach a goal—not on a goal that is poor. (Saving the environment is a noble goal. To, like Germany, make the abolishment of nuclear power the first priority on that road is idiotic, and shows a horrifying ignorance of the relative benefits and dangers of nuclear power, be it in absolute terms or compared with fossil fuels.) Similarly, there might be disagreement over what price is worth paying to achieve a goal. (Some see virtually any price as acceptable to reach “net zero”; others emphatically do not. More interesting questions for this text include what costs might have been unforeseen or much larger than predicted. For instance, politicians in a great many countries appear to have severely underestimated the effects of “net zero” policies on the economy.)

A partial solution

My suggested partial solution is exactly evidence-based politics:

With any measure instituted, politicians must be able to show:

1. That there is sufficiently strong evidence that the measure will actually achieve the intended goal in a manner sufficiently effective, efficient, timely, or whatever might apply to the case at hand.

   Exactly what this evidence should be will vary from case to case. Solid scientific proof is the ideal, but will rarely be available. In many cases, historical experiences that point to a strong likelihood of the intended outcomes in their sum (not after cherry-picking) will be an alternative. Ditto e.g. the sum of opinions of experts (with some additional caveats, e.g. that the experts must be politically and ideologically neutral, must not stand to gain from the measure, must not work for a special-interest organization.) Other options yet might exist.

   Under no circumstances are unsubstantiated claims that “science says” enough. Ditto, drawing on a field with ideological problems (notably, in the social sciences) while ignoring what other scientific fields say. Ditto, reliance on a small number of individual experts (or those that fail the above caveats). Etc.

   ---

   Side-note:

   Concerning “science says”:

   What science does say is a valid argument. However, the mere claim that “science says” is not. Those willing to dishonestly claim “X” can claim “science says X” with equal ease.

   ---

   The more important the issue and the more far-going the potential consequences of the measure, the more important it is to have strong evidence.

   (The following items are subject to similar remarks, even absent a restatement.)

2. That costs, risks, potential side-effects, and similar, have all been sufficiently analyzed and given sufficient weight in the decision-making process.

   This should, when at all possible, include a thorough and publicly available cost–benefit analysis. (Also see excursion.) Indeed, one of my main complaints against the COVID-countermeasure era is that no decision maker actually seemed to have made such an analysis—and if one has been made, it has, contrary to the principles of evidence-based politics, been kept out of the public eye.

   A particularly common problem is that incentives are not considered, even when they can cause a measure to fail. For instance, a tax increase on the rich can cause the rich to move abroad, which will cause disadvantages, including a reduction in the tax base and less local purchasing power. For instance, giving subsidies to buyers of electric cars is likely to do more to increase the price of electric cars than to make them more affordable. Both examples might seem obvious, but have often been neglected in political decision making.

3. That the decision-making has been sufficiently holistic.

   For instance, the virtually exclusive focus on medical issues around COVID brought on some horrifyingly poor decisions, with consequences including, among many others, a severe damage to the economy, individual bankruptcies, and loss of education among students. To make matters worse, the extensive effects outside medicine have had considerable indirect effects on the health of the people, e.g. in that economic problems can take a physical toll on someone—and here we talk millions upon millions of such someones. It might even be argued that there was a virtually exclusive focus on COVID over even other medical issues, as even more directly medical problems were not given due weight, e.g. in that many health problems were only investigated with a great delay, with consequences like potentially treatable cancers not receiving treatment in time.

In later stages:

1. The actual effects of various measures over various timespans must be investigated, documented, and compared with the expectations.

   ---

   Side-note:

   Here it would be helpfully if the measures came with quantified predictions, which would make a later comparison of relative success and failure easier, but this, I suspect, would only rarely be realistic on more than a ballpark basis. An interesting thought, however, is how being forced to give even a ballpark quantification could affect the behaviors of politicians.

   ---

2. The various measures must be re-evaluated in light of these results and, when relevant, be adapted or reversed.

The solution perverted

Unfortunately, there is reason to fear that evidence-based politics might be corrupted or circumvented.

For instance, Leftist politicians have a long history of drawing on pseudo-sciences in preference of proper sciences, e.g. in that they prefer Lysenko to Darwin or gender studies to biology and psychometrics.

In some cases, e.g. with Keynes, there is reason to suspect an (at least partial) reverse causation: Politicians have not used Keynesian policies because he was the messiah of economics. They have used Keynes as an excuse to implement the policies that they already wanted to implement, or to justify the policies that they already had implemented, anointing him as that messiah in the process.

In a bigger picture, even the harder sciences are increasingly under ideological attack and stand a risk of being corrupted. (While the social sciences have been in problems since deep in the 20th century.) There are research topics or results that are considered taboo, “diversity” increasingly trumps scientific merit when it comes to appointments, even scientists are increasingly forced to sign political declarations (notably, again, concerning “diversity”), etc.

Excursion on general attitudes and approaches

Free debate, transparency, etc.

For this to work, attention must be paid to the surrounding intellectual climate and circumstances. For instance, it must be possible to freely debate the pros and cons of any suggested or implemented policy, and to do so without artificial censorship, without being shouted down by government propaganda, without being exposed to the type of anti-factual “fact checking” and wild ad hominem that was so prevalent during the COVID-countermeasure era and is so preferred by the modern Left, etc.

Similarly, politicians and governments must be sufficiently transparent about their goals, approaches, argumentation, whatnot, that these can actually be independently investigated and criticized. Equally, the results of later investigations must not just be put in storage with a “top secret” stamped on the file—they must be readily available to any and all interested parties, including scientists, journalists, and curious or critical citizens with no particular credentials or accreditations. (Possibly, with exceptions for findings that truly might require secrecy.)

Politicians with a scientific mindset

It would also be immensely valuable, if the typical politician had a more scientific mindset, especially through being more set on finding out the truth than on insisting that an existing position, no matter how weakly supported, is the truth.

However, achieving this would likely require great changes to e.g. the regulations of who is at all eligible to be elected, and it would be very hard to actually test for such a mindset.

To boot, many actual scientists lack a scientific mindset, which makes any hopes of extending the mindset to politicians highly optimistic.

As an aside, the fact that Angela Merkel had a background as a scientist was what, in my eyes, originally seemed to make her a great choice for chancellor. She turned out to be an enormous disappointment, a poor decision maker, and one that failed even where I would have expected specifically a scientist to act radically differently (notably, with regard to the abolishment of nuclear power and her apparently complete lack of skepticism to exaggerated COVID-scenarios.)

Excursion on performance evaluation, liability/consequences, etc.

With evidence-based politics, it would, to some degree, be possible to evaluate the work of the politicians, e.g. to point out that the implemented measures suggested by A fared worse than those by B or that those by C almost always had a negative net-result. This would give a far better indication of who is or is not worthy of re-election than the often vague impressions available today—not to mention sheer superficialities, e.g. who visited a pet zoo and who was caught in an extramarital affair.

Legal consequences and/or liability could also be on the table in a different manner than today, which could prove a very welcome check on the excesses and serial mistakes of some politicians. (Today, the only real check on politicians is the possibility that a screw-up reduces their popularity too much, leading to election losses.) This, especially, when a (failed) measure was implemented against clear warnings and a manifest risk that it would do more harm than good, as well as when no corrective actions were taken once the post-implementation problems were documented.

The exact form of such consequences and liability is another matter, which I leave entirely open, as there are a great many factors to consider. For instance, even something as basic as the right “unit” can be tricky— should the focus be on the individual politicians, on political parties, or on some other grouping? (With answers that can vary from case to case.) For instance, too harsh or too easily triggered consequences might create such fear of failure that they do not just prevent many of the current idiocies but also cause inactivity among politicians even when and where they should be active. (This raises yet another question, namely how “negligently inactive” politicians should be handled, be it in general or within specifically the scope of evidence-based politics, which might make an interesting extension at a later date.) For instance, there is a risk that political opponents would try to abuse the system for bad-faith attempts to hurt or eliminate political opponents.

Effective restitution for damage done will rarely be an option, unfortunately, as the amounts collectable will almost always be drops in the ocean. (And how would restitution be fairly divided among the injured parties?)

Independent of evidence-based politics, an interesting thought is that parties or politicians might be forced to name some certain targets that must be reached (should they be elected, should a certain measure be passed, or similar), e.g. a minimal level of yearly growth and a maximal level of yearly inflation, with consequences ensuing if these criteria are not met. (Here, too, I leave the details open. The targets that are relevant will obviously depend on the position/measure/whatnot at hand—the local mayor can do little about national inflation, e.g.)

Excursion on the precautionary principle

In a similar vein, it would be welcome if politicians took greater heed of the precautionary principle: When decisions are made for an entire people (and, potentially, its descendants), great caution is needed.

COVID is, again, a good example and, sadly, one where the precautionary principle was grossly misapplied: A sane approach would have demanded that the precautionary principle was applied to the countermeasures, to the changes instituted by governments everywhere, to the deviations from the status quo, whatnot. This, especially, as these countermeasures (etc.) had little or no precedent, were extremely far-going, and were applied more-or-less worldwide. Instead, the “reasoning” was that “we do not know how bad COVID might be—so we have to clamp down as hard as we can, just in case”. This despite that there was no true reason to suspect a “The Stand” situation; despite the largely false alarms around SARS, swine flu, and whatnot in the recent past; despite how the world had survived e.g. the Spanish Flu with far lesser medical knowledge and resources; etc. (And what did more damage? The Spanish Flu or the Great Depression?)

In particular: The greater the uncertainties about outcomes and the higher the stake are, the greater the importance of the precautionary principle—and the more important it is to apply it correctly.

An interesting difference between businesses and governments is that businesses tend to test whether something works first and adopt it in large scale only if the tests are sufficiently successful, while governments tend to leap right in with the large scale, despite the stakes being so much higher. In the overlap between scientific spirit and precautionary principle, why not make more experiments? If something can be tested in a limited geographic area (not everything can), do so before going national. If a radical increase of something, beyond historical norms, is suggested, change it a little today, see what happens, and then consider increasing it more tomorrow. Etc.

Excursion on existing quality measures

In a slightly earlier version of the preceding excursion, I derisively noted:

(Well, there is often some type of committee at work before the leap is taken, but a committee does not replace an actual test.)

Within half an hour, I grew a poor conscience, because there is often quite a lot more going on in terms of what might, for want of a better phrase, be called “quality measures”. Moreover, this “more” should be properly contrasted with some of the main discussion above.

Now, what happens in any given country, on any given level of government, at any given time, etc., will vary considerably, but some important examples (without any claim of completeness) follow below.

Of course, often, there is no guarantee that they actually take place, that they are taken seriously enough, that negative feedback is not swept under the carpet, that they are not abused as alibi-measures to justify the wanted decision, etc.—and the degree of transparency of both the process and the results might not reach the level needed for evidence-based politics.

Most importantly, maybe, they do not apply everywhere. For instance, a law might be preceded by a parliamentary debate, committee work, etc., but a big foreign-policy decision that does not require a law (or some other form of explicit parliamentary approval) might be made in a very small and private circle.

1. Parliamentary debates, speeches, etc.:

   While valuable in theory, the practical value is disputable, as voting is usually dictated by the party line and as, when not, preconceived opinions tend to dominate. It might give the one side a chance to present arguments that the other side will largely ignore but might look good in the press, allow some mutual pontification, fulfill some formality needed to allow a vote, or similar. (In some cases, notably the U.S. congress, variations on the filibuster might be relevant.)

   At the end of the day, they consist of politicians debating each other and whatever value is present has little to do with evidence-based politics.

2. The aforementioned committees:

   These are typically made up of politicians or those “politician adjacent”, which is a bad sign in terms of getting impartial feedback, feedback representing a wide variety of relevant views, feedback based on a deep understanding of the domain at hand, etc.

3. Various consultations with potential stakeholders and domain experts:

   These are a strong step in the right direction, but the choices are usually too incomplete and/or poor. For instance, unions, guilds, and the like might be included not just as stakeholders but as experts. These might well have some expertise, but they are ultimately special interest organizations that are too partial and too likely to give “expert advice” that favors their own causes. (Note e.g. the German IHKs and the problems that they cause. TODO import from Wordpress and link.)

   To boot, as with many organizations, the domain competence at the top tends to be weak, as the organizations are run by managers and not experts, as requests for feedback might end up with the PR (or whatnot) department, and as many have a dearth of truly intelligent individuals in the organization. (A plumbers’ union, e.g., is likely to have a far lower average I.Q. than a physicians’ union—should one exist. Another complication is that those of low competence are more likely to be unionized in massively large organizations and, thereby, have an indirect political clout that those of higher competence lack.)

   Likewise, the influence of managers often comes at the cost of those who should truly be consulted, be it as stakeholders or as experts. If, e.g., a university is consulted on topics of education or research, what are the relative chances that the feedback comes from professors respectively administrators? In an over-simplified (maybe, caricatured) example, assume that a “Why have our universities fallen behind in student accomplishment?” is sent to a principal. If this is sent on to a few fact-focused professors, they might complain about e.g. a drop in student quality, diligence, responsibility, dedication, or similar; if to a few woke administrators, the answer might be “Too little diversity!!!”. In the first case, a focus might be put on increasing student quality, e.g. through increasing the influence of academic merit on admissions; in the second, on increasing the importance of belonging to the right societal group at the cost of merit.

   Then there is the issue of whether and to what degree any given organization favors the goals of the members of the organization, the leaders of the organization, or, even, the organization (as such).

   ---

   Side-note:

   My own experiences as short-term member of several organizations have been less than stellar. (Note e.g. an older text; in particular, with regard to the German professional organization VDI, which disingenuously claims to represent German engineers.)

   ---

   Many special-interest organizations (especially, when of a political nature) have a strongly biased, sometimes outright fanatical take, and are highly unsuitable for consultation—even when they proclaim themselves to be experts or demand to be consulted.

   Individual experts, in turn, are rarely sufficiently widely chosen, “resident experts” are disproportionally likely to be called in (even while having a strong risk of partiality), members of pseudo-scientific branches of academia (e.g. gender studies) are all to often given an ear, and there is a risk that the experts are chosen based on the likelihood that they say what the politicians want to hear.

4. Polling of voters (party members, whatnot) for their preferences:

   Voters typical lack expertise and a sufficiently holistic view, understand even less than politicians about e.g. economics, often base their opinions on what the politicians tell them to believe, or are otherwise poor sources of feedback from a viewpoint of evidence-based politics.

   ---

   Side-note:

   Such polling might be legitimate for other reasons, however, e.g. to ensure that the politicians do not lose contact with the broad masses.

   Even then they can be problematic, as the motivation is often the wrong one—to find out “How do we manage to be re-elected?”, “With how much can we get away?”, and similar, instead of “What does the people want?”.

   ---

   At an extreme, we have the Swedish plebiscites, which are advisory. (While plebiscites that are binding are a different beast altogether, notwithstanding a similar risk of manipulation.) On some very few occasions, a question has been put in front of the people for a full poll. Of particular interest to this item is the 1980 plebiscite on nuclear power, where both proponents and opponents were failed at different times, as (a) the voters were given a three-fold choice between abolishing, abolishing, and abolishing, with only a variation in modalities and no option of keeping nuclear power, (b) nuclear power has still, in 2024, not been abolished. (A more detailed text will be published in due time. TODO link once published.)

Excursion on cost–benefit analyses

A cost–benefit analysis is, obviously, easier when everything can be brought down to some common measure, e.g. by how much the bottom line of a business would increase or decrease after it takes some measure.

This is not always possible or sensible even with businesses, and it can be much harder in politics, because oranges have to measured against apples—and grapes, figs, melons, ...

Moreover, the uncertainties about outcomes, probabilities, expectation values, etc., are often quite large.

This, I would argue, makes it the more important to go as far as the numbers allow and to allow others to make their own evaluations of the data and calculations. For instance, if a politician sees a gain of three apples and a loss of two oranges, this might be perfectly fine according to his priorities, preferences, whatnot, but there is no guarantee that others see it in the same way, and by forcing the calculations into public knowledge, those with different priorities have a better chance of making themselves heard, of gaining a sufficient understanding of what politicians to prefer and avoid, etc. For instance, if politicians are forced to publicize their probability estimates (or their failure to make such estimates!), others can apply their own judgment, repeat the calculations with own estimates, etc.

(Note how strongly the above applies to the COVID-countermeasure era and the extremely poor, top down, single priority, whatnot, decision making around COVID.)

I would even argue that great problems with the construction of a cost–benefit analysis is an indication that a certain measure should be avoided entirely, that a decision about the measure should be postponed until a better analysis is possible, or, at a minimum, that the measure should be approached with unusually large caution. A special advantage is that such avoidance/postponement makes it harder for politicians to push decisions based more on personal preference or emotion than reason. (Note e.g. sloganeering like “Even one X is one too many!” or “No cost is too high to achieve Y!”. These might sound good, but if members of the public learn that the cost of achieving an additional apple is thrice as many oranges than they might otherwise have assumed, they will be less enthusiastic and less likely to be fooled.)

Excursion on other medical ideas

There are at least three other “medical” ideas, with intended companion texts, that overlap with the above: informed consent, the motto of “above all, do no harm”, and “minimal invasiveness”. I do not seem to find the time for these texts, so, at the risk of going off topic, some preliminary remarks:

The citizens of today are often not given a true option to consent or not consent, e.g. because important decisions are made in an un- or pseudo-democratic manner. (Where it is understood that declining to consent makes someone exempt from whatever the consent was required for. This as opposed to e.g. a Lauterbachian “you will be punished until you do consent”.) When they do have an option, the consent is hardly ever informed, because crucial information is either not made public (politician’s fault) or not digested (citizen’s fault). Here a different take is needed, where the option of informed consent is provided and those who have voluntarily not digested relevant information abstain from e.g. voting. (Other complications include if and when the consent of the people as a whole should be enough and when that of the individual citizen is needed and how consent should be given, e.g. whether a vote is enough—and, if so, how to handle those without a right to vote. My standing recommendation is to minimize collective decisions that affect the individual in a non-trivial manner to the degree possible, as well as to increase individual consent.)

Politicians are far too keen at meddling in a manner that does harm, be it out of ignorance, lack of due concern, or for some other reason yet. This in at least two variation: Firstly, that poorly chosen acts lead to an overall net damage (by some measure). Secondly, in that harm to the one is accepted in order to favor the other—or, worse, to favor the government (the politicians, whatnot). While neither of the two variations can be entirely avoided (if in doubt, mistakes do happen), politicians must approach their job with the intent to minimize them to the degree possible and at all practical. Further, to re-invoke the precautionary principle, an awareness of the risk of unforeseen harm must be present and politicians should err on the side of not meddling with an eye at that risk.

Policy and laws are far too often overly invasive, do more than is needed to achieve a particular goal (and with the surplus often being a negative), intrude on matters that should be kept private and/or left to the prerogative of the individual citizen, or similar. Much good could be achieved by cutting down on this. Also note the idea of a “night-watchman state” and some thoughts on “good”, “bad”, “minimal invasive”, whatnot, laws in another text.