Ignoring feedback as the root of most modern problems

I might sound as a sidekick to Captain Obvious but some things are needed to be said again and again.

Life in its entirety is built on a feedback loop: if you react on external conditions to maximise positive output you’ll survive; it’s not just evolution, individual organism also would die if they don’t search for food when they’re hungry and don’t flee from danger. But when you have large systems with some resilience, some of the feedback may be ignored for some time. And sometimes all of the feedback gets ignored permanently and that is the start of that system downfall. It has happened before the humanity (dinosaur extinction most likely happened because they were unable to adapt to the changing conditions—because they could ignore the changes until it was too late so they were mostly replaced by more agile animals) but here we’ll review human-made systems (mostly social and political) and failures of those.

Essentially, the cycle works like this: at first a system emerges (or is created) to fulfil some purpose, it grows large, then it starts getting optimised for a target different from the initial one, and when the differences between its initial purpose and the goal it serves grow large enough for that system to face an uneasy choice—get reformed or get eliminated. This happens because the feedback from the wrong decisions at stage three (optimising for a different target) get ignored. So let’s start with reviewing how this happens.

You probably have heard about Goodhart’s law (and if not, it’s a shame) which essentially says that if you optimise something only for a certain metric, that metric will stop mattering. To give a technical example from the area I know: there are various ways to calculate how close lossy-encoded video is to the source and the higher the value of such metric the better video should look to human eye. So far so good, right? But every metric can be manipulated and some researches found that by applying some preprocessing the popular VMAF metric can be improved while the picture quality may feel degraded (it’s not purely theoretical, here in AV1 encoder source there’s some code based on their earlier ideas). Also I heard an anecdote about Korean researchers making an AAC encoder based on neural networks that gave excellent PSNR while sounding like garbage to human ear. Now that you have an idea on how it works, let’s move to less quantifiable things.

I can’t say I understand modern economics well, but essentially in the old days companies were supposed to bring profit by providing a product or service better than the competition (in the best case by being a monopoly). Then the companies became too large to be owned by one guy or family (there are exceptions of course) so the goal became to increase the worth of the stock that shareholders possess—by increasing its market value. Again, initially such stock prices were tied to the easily quantifiable things like assets, income and profit; now it’s more about expectations of growth. Somebody from the past simply would not understand how a company having losses for years still has its stock growing. My cynical view is that it works as long as there are enough people willing to believe that it works (which is also the case with modern economics in general IMO).

Now let’s look at the forms of government. Usually there are no clear goals in either of the systems and people often can’t predict the consequences of their actions (and when they can, they’re not willing to). And there is the usual human tendency to make things better for oneself no matter what consequences it bears for the others. Thus in order to keep system vital you need to get feedback from the people (even if it’s just to know where to send forces to squash rebellion like in today Iran). That’s why China keeps several systems (party members, state security and the state news agency) to report stuff from the regions to the centre. That’s why Winston Churchill said that “democracy is the worst form of government except for all those other forms that have been tried from time to time”. Voting in sane(r) countries is a way to keep the authorities informed about what people want (even if the results of the voting are ignored afterwards). IMO the system could be improved by having a test on the basics (e.g. what parties are out there, what are their governing principles and key points of the programs—just to make sure you make a conscious vote) and account votes of those who passed the test and not separately (in order to do what the concise people voted for and if they’re indecisive—do what the majority wants). But of course such system can be exploited (plus it’s somewhat harder to implement) and we have Gresham’s law in action here (a voter is a voter and any populist would prefer a thousand people voting for him out of tradition to ten people voting because he served their interests but who might not vote for him the next time). The cases of what happens when feedback is ignored will be reviewed a bit later.

Now let’s have a look at social networks. From what I see, originally they were created to attract people by connecting them based on some feature (and make money off them of course). With the time though the goal has mutated to attracting advertisers and selling them audience based on some features. I’m not going to describe what tricks are used to make users behave in the way advertisers like the best but it seems to turn users off those platforms.

As we got to the point where not taking feedback into account has led us to a crisis, there are three possible outcomes: keep ignoring the feedback (aka Soviet Army approach), try to manipulate the metric in hope that it will improve things (shamanism) or do something to the reality to make it conform to the metric (Froogle approach). If you think that taking feedback into account is a possible option too you are an incurable optimist.

Ignoring feedback altogether is a rather simple thing to do. If you don’t like something just call it fake news or enemy propaganda and be done with it. Why I called it Soviet Army approach? Because for me it’s associated with the tradition that was common for the commanders in russian empire, USSR and current russia: keep doing what’s ordered no matter how pointless it is. There are jokes about e.g. painting grass green and asphalt in dark grey to conform the rules, but considering all the variations of it I’ve read in the memoirs and heard from people who served in Soviet Army I doubt it’s merely jokes. Similarly we can see how in this war russians are attacking Bakhmut (or Sieverodonetsk earlier) and how they made Chornobaivka a legend. Denying reality just prolongs the inevitable.

Manipulating metrics is another step, where you hope that by cooking the books and rigging the process you’ll get the desired outcome. As Richard Feynman said, nature cannot be fooled—but what about people? Maybe if we report that the majority supports us the majority will think they’re supporting us because everybody else does? This starts with rigging elections and ends with russia where the only feedback taken into account is what the führer wants to hear, so the election results are always falsified to have the ruling party get the “convincing” majority of votes, opinion polls are treated by population as “the authorities sent me to ask if you’re satisfied with everything, yes or yes?” and it is widely speculated that only good news about the war get to the führer-in-the-bunker (unless they’re too hard to hide, then they’re greatly delayed as it happened with their cruiser, fleeing Kharkiv region or Kherson). In theory this approach may work in some cases where you act in a system powered mostly by human beliefs and you have factual information yourself, otherwise it causes more damage than the previous case.

And finally we have the case where you not merely reject feedback from reality but rather try to force reality to give the feedback you want. I called it Froogle approach after one story (sorry, can’t find the link): when this search service for products on sale was still in development, the team encountered a case where a certain irrelevant product (a garden gnome) was showing up in the results for completely different things. No matter how they tuned their algorithms, it kept showing up—until one day one guy from the team went and bought it thus removing it from the search results. This is also reminding me of the final chapter of The History of One Town by Saltykov-Shchedrin where a satirical depiction of Arakcheyev sees a river on plains that does not fit into his world view and decides to remove it by filling it with rubble from the demolished town (he wanted to build a military-agricultural colony on its place anyway). As you can imagine, it worked for about one day and then the river found its way again. The famous whipping the sea by Xerxes the Great probably belongs here as well. Depending on the resources you have (and against what you fight) it may work out or not—but it’s fun to watch from afar (from close distance though it’s a horror).

We learn from the mistakes (preferably not made by us) and feedback is a very important part of that learning. Pain tells you that what you’re doing is hurting you and you’d better stop doing it, that’s how children learn not to grab hot objects and pointy things. Sadly nowadays in many cases you don’t suffer from the consequences of your mistakes (because the society values individuals a bit too much to allow that) and in some cases it leads to tragic results. Even worse when it happens to the whole countries (because Wesphalian sovereignty is an international counterpart to humanism).

In 2008 russia started a war on Georgia and what was the reaction? The famous EU report that stated essentially that the conflict happened and they don’t know the reasons behind it and can’t verify the claims of participating parties. A year later Obama government started with resetting the relationships. As the result it empowered the aggressor to do more of the same in the future (1938 called and sent its greetings from Munich).

In 2014 russia started essentially the Third World War (I wrote on this opinion of mine) by occupying Crimea and fighting in the East Ukraine using puppet “people republics” (like soviet russia did in 1917-1922). What was the reaction? Essentially it boiled down to “bad russia, don’t do it again—and you, Ukraine, should reconcile” (plus a couple of rather formal sanctions to show some concern). And in the same time russia tried to use its gas as a leverage (not for the first time though and not for the last). Baltic states took the hint and made their systems independent from russia: there were LNG terminals built and new legislation for controlling gas suppliers was passed (like requiring minimum storage fill level). Germany did nothing and as the result it met the second phase of war with near-empty storages (belonging to russia) and no means to transport gas but the pipelines from russia (mostly owned by guess which country)—all against the Directive 2009/73/EC. To me this alone is enough to start investigating Merkel (and preceding) government for high treason, but we all know this is unlikely. The lack of punishment for acting against national interests and in favour of another country will encourage others to do the same (see Scholz and China). And there’s still an issue of Hungary and its well-paid affection to russia…

The situation in the USA is not much better as both parties do not want to recognize that people might have a reason to vote for their opponents and it’s not just russia/China/big tech meddling with the elections. And instead of thinking what can be done to win more voters the parties mostly pass the blame and try to turn their views into a small religion. Also I find the calls to disregard the laws because the other party “cheated” rather disturbing. Thanks to this other countries see America as a laughing stock and, more importantly, withering power and expect to occupy Ukraine, Taiwan or Syria without repercussions.

I want to end this rant with a passage from Going Postal that resonates with me:

You see, I believe in freedom, Mr Lipwig. Not many people do, although they will of course protest otherwise. And no practical definition of freedom would be complete without the freedom to take the consequences. Indeed, it is the freedom upon which all the others are based.

Comments are closed.