On Chesterton's Fence

post by Martin Sustrik (sustrik) · 2021-04-29T14:56:33.747Z · LW · GW · 9 comments

Let's think of the evolution as a conservative force for a bit.

The common narrative of evolution is that it is that great force of progress, optimizing the organisms, casting away inefficiencies, multiplying the species, and generally, driving living things to become more complex, starting from the simplest possible lifeforms, such as protozoans, up to that pinnacle of progress that is ourselves, the vertebrates.

But consider a plant that happens to have a mechanism to survive droughts, such as ability to store water in the roots. If there is no drought this year, that mechanism is a dead weight. It is costly to maintain and there's no immediate benefit form it. The plant would be better of without it. But the next year there's a drought and the plant survives. If the evolution was super progressive, the plant would optimize out its water-accumulating capacity in the first year and then die in the second year leaving no progeny behind.

A conservative may thus say: Evolution is a way to store knowledge of the past. It's a trove of mechanisms that have once proved useful for surviving. "Optimizing" it is just a nice way of saying that the instruments from that war chest are being thrown away.

And, of course, neither our strawman progressive or our strawman conservative are considering the full picture. Organisms benefit from having inherited the survival tricks from the past but also from being able to respond quickly and in novel ways to the new challenges. (But, to be fair, the conservative view is much easier to argue for. Arguing for the progressive view tends to lead to the theories on group selection, if not to some semi-mystical élan évolutif.)

In the end it's all probably just a race with time. The events that are frequent enough (the low temperature during the night, the cold in the winter) are worth optimizing for. Infrequent events (the ice ages) not much so. In the former case the tools for dealing with the event are worth keeping in the war chest. In the latter case they will be inevitably thrown away and we have to rely on the fact that at least some organisms will be, for random reasons, better suited to withstand cold and that those organisms will be the lucky winners of the evolutionary race when the next ice age comes.

To switch from evolutionary biology to political science, let's recall what Jean Monet once said on the topic of creating common European institutions:

The tragic events we have lived through and are still witnessing may have made us wiser. But men pass away; others will take our place. We cannot bequeath them our personal experience. That will die with us. But we can leave them institutions. The life of institutions is longer than that of men: if they are well built, they can accumulate and hand on the wisdom of succeeding generations.

It is easy to notice that this is the conservative argument as stated above. Institutions are our war chest of tools designed to deal with the problems that we've encountered in the past. But they don't necessarily give us the ability to deal with the novel challenges. To do that we need to create new institutions and add them to the chest. And, similarly, if the problem that lead to the creation of an institution doesn't occur for a sufficiently long time, the institution will eventually erode, get removed or repurposed and won't save us when the problem hits again.

And while this article doesn't offer any particularly new ideas, it may be worth thinking about before removing a fence.

9 comments

Comments sorted by top scores.

comment by kithpendragon · 2021-04-29T17:24:21.677Z · LW(p) · GW(p)

Thoughts off the top of my head:

The primary problem with Chesterton's Fence is that it doesn't come with enough documentation! With the purpose written in big block letters right on the fence, we might be able to create enough of an entropy pump to cause a rare-use institution to remain "fit for purpose" when it becomes needed again.

All too often we see agencies and organizations built for the short term, funded by government will alone and with only vague (but, in the moment, entirely understandable) directives that lead to a situation of reduced usefulness when their monotask is perceived as "completed". It seems to me that there could be much more value in being more explicit than we typically are when creating these institutions, and in creating them with the long-term in mind.

Many established institutions have some brief charter or briefer mission statement designed to inform their purposes. But that might not be enough to let the progeny eventually figure out what to do, or to refrain from eroding or repurposing those projects as their need becomes less acute. We might do better if, along with the standard purpose-declarations, we make a habit of including some additional components.

From the outset, there needs to be a comprehensive history explaining why the institution was founded. This should include records of relevant mass media, associated personal communications and other private documents, and a report explaining why each was chosen and how the founders were thinking about it at the time. This would provide for anybody who cared to look, a comprehensive record of the context of the creation of the institution.

Too, we know that institutions tend to degrade and change over time as the resources required to maintain their functioning become "needed" for "more important" things. With this in mind, we should be able to create systems that operate on the least resources possible, but are designed to serve a basic function during normal times and expand quickly in an emergency. For example: such an institution might be tasked for most of the time with stockpiling supplies, and with maintaining a distribution system for those supplies and a network of professionals who would be prepared to spring into action when needed (including such resources, education, and training as necessary). Perhaps it would serve in "normal" times as the primary provider of these resources in order to keep stock rotating appropriately, and as a think tank for emergency protocols. Maybe it would even oversee a certain amount of the education for the field it services. This would be done at market rates to provide capital, and would also help to maintain the ready network of professionals.

Also, we should pre-commit to deferring to our emergency plans when emergencies actually occur. What good is a well-researched carefully-planned protocol if half the population is thinking "well... that's kind of inconvenient. What if I just don't?" We need regulations "with teeth" in place before the emergency just like we need good contingency plans to follow in the first place.

Replies from: Pattern, Pattern
comment by Pattern · 2021-04-30T05:14:40.583Z · LW(p) · GW(p)
and a report

This might not be enough in practice for everything. Talking to an actual person who lived through X, and why they made the decisions they did, might be necessary. (As someone who doesn't appreciate everything turning into a podcast instead of being text, I'm not the biggest fan of this. That said, I think it's getting cheaper to video record everything (though liability/stuff that operates via the same mechanisms might work against this). Condensing and summarizing lots of video takes work, but can sometimes lead to new insight. (As can asking experts.)


Also, we should pre-commit to deferring to our emergency plans when emergencies actually occur. What good is a well-researched carefully-planned protocol if half the population is thinking "well... that's kind of inconvenient. What if I just don't?" We need regulations "with teeth" in place before the emergency just like we need good contingency plans to follow in the first place.

An other approach is get people to come to an agreement. This may take time, but if the emergency doesn't happen often, you may have time. (Those involved in resolving the agreement may wish there weren't so much time.)

Also, this might be something already addressed but - disaster plans might have to change when another disaster occurs in the middle of that other one, if there aren't already plans for that. For example, people might stop committing crimes during natural disasters, but viruses probably won't go on vacation.

Replies from: kithpendragon
comment by kithpendragon · 2021-04-30T19:36:26.539Z · LW(p) · GW(p)

Talking to an actual person who lived through X, and why they made the decisions they did, might be necessary.

Sounds good! Add all the interviews to the report. The more original sources we can collect in the moment when we feel the need to create an institution, the more people will eventually have to go on when they're making decisions about these institutions in the distant future.

An other approach is get people to come to an agreement.

Yes; and in the generational timescales we sometimes get to work with between certain emergencies I think this can be solved by education. At least the fundamentals need to be absorbed into the culture in some way.

disaster plans might have to change when another disaster occurs in the middle of that other one, if there aren't already plans for that

I hope we have enough time between instances of the same kind of emergency to think of those contingencies too. Ideally, our emergency-preparedness institutions should all be able to talk to each other and coordinate their plans together.

comment by Pattern · 2021-04-30T05:12:48.653Z · LW(p) · GW(p)

I don't disagree about the importance of documentation. That said:

In order for an approach to work, does the explanation of the approach have to be correct?

Replies from: kithpendragon
comment by kithpendragon · 2021-04-30T19:43:13.322Z · LW(p) · GW(p)

Since we would be designing institutions to last at least a few generations before their full functionality is probably needed, I hope we would be able to use the meanwhile to work toward more correct approaches with more correct explanations to back them up. That said, if some procedure kicks in during an emergency, and it happens to save lives despite the reasoning behind it (or the explanation of that reasoning) turns out to be incorrect, it's still a win vs doing nothing! Then we can learn from the new instance of that emergency and do even better next time.

comment by Pattern · 2021-04-30T04:49:48.557Z · LW(p) · GW(p)
conservative
progressive

Neither is obviously a good description. More raw approaches using terminology that stem from observation:

  • Survival of the fittest
  • Survival of the flattest

May fit better, though it may take a while to find what you are looking for, or longer to explain. 'Evolution favors 'robust' systems that don't get blown away because they harvest too much power and are unable to adjust for variance which is amplified under maximum extraction, which is why you can edit genes to make photosynthesis more efficient, and it may even work well for edited plants in a laboratory under grow lights where clouds don't pass across the light source, and winds don't blow around sources of shade high above for many levels.

Don't die because you absorb too much power is a problem, and evolution*** solves it.**'


Intuitively, preparing for an ice age, probably works against preparing for the opposite (from an evolutionary perspective), most of the time.


eventually erode, get removed or repurposed

And this is an example of out how, a piece that isn't about evolution, ignores things because they aren't relevant. For example, viruses can sit around, biding their time, rather than having to 'go on living, and carry on via generations'. If you want to talk about 'living knowledge' then 'how elephants deal with drought' will be more specific and related. (If a herd has members old enough to remember what to do*, like where to go, then the herd of elephants is a lot more likely to survive.)

There's also stealing genes, which would seem to have some of the same problems that trying to copy institutions would have, and yet somehow works? Might be risky, but when failures die out...evolution is like survivorship bias turned up to eleven. And it works.


*Has to have been born, and be old enough to pay attention and remember.


**

arguing for the progressive view tends to lead to the theories on group selection,

Factors keeping ecosystems balanced under adverse conditions seem to involve groups, though larger and less species focused ones than this usually seems to refer to, in common usage. The frequency of symbiotic relationships (for better or for worse), seems suggestive of outside species relationships. Perhaps a result of a lack of competition, or increased ability to synergies when both parties are able to evolve to do different things.


***Brute force, an approach renowned for its elegance.

comment by Slider · 2021-04-29T18:35:55.946Z · LW(p) · GW(p)

My mind matches this to be the conflict between the magic the gathering colors of red and green. Green likes accumulation and red likes reactivity. Preposed sideways solution is the white option, provide an eternal order. However ther are som downsides to this. White likes to sanctify and guard but it doesn't actually like to add. Green would fill up the chest and while white is better at ensuring no loss of progress it has equal problems of letting stuff in than strength in preventing stuff from leaving. Yeah your valuables are safe from any thief but you can't make any deposits. (The filpside of "eternal" as "incorruptible" is perfect stagnation) 

comment by ACrackedPot · 2021-04-29T16:36:00.902Z · LW(p) · GW(p)

The evolution-institution framing is a useful one, and the metaphor, I think, extends further than that.

Not all evolutionary mechanisms are costly, in a certain sense - the major maintenance cost is informational.  If an evolved mechanism doesn't get used, it doesn't necessarily get selected against, but it also doesn't get selected for; the mechanism can be eroded by informational degradation / entropy.  Or it can be repurposed for some new issue.

The same is true of institutions, and we can observe it happening; we should expect chemical disasters every so often, just because the regulatory apparatus that ensures that chemical disasters don't occur, is renewed only when they do occur.  Funding is one mechanism for this - maybe funding never decreases, it just doesn't keep up, until a disaster occurs.  But also institutional drift - without chemical disasters to keep the public eye on the regulatory institution, the institution's purpose is gradually going to drift away from preventing chemical disasters, and drift towards the interests of the people running it (ensuring they have comfortable jobs, for instance), as one form of repurpose - but it could be repurposed to deal with some new issue as well.

Which is to say, the original adaptation - or the institution - could be serving another purpose entirely when it is needed again, and not actually be able to fulfill the requirements for which it was originally developed.  More, the new purpose could be important enough that you can't just switch back to the original purpose.

One ramification is that Chesterton's Fence doesn't necessarily apply, because the thing that was put in the war chest, by the time you need it again, may not even be in there anyways.  More, there may be pieces that are maladaptive to the new purpose, since the structure was created to fulfill a different purpose.

Replies from: Pattern
comment by Pattern · 2021-04-30T05:25:51.632Z · LW(p) · GW(p)

So an institution for dealing with say, earthquakes, stays on track by repurposing itself* for a new/future earthquake, whenever one comes along, and that's how it stays on track?

*More evolving than predicting the future, because predicting the future is hard.