Hindsight Analysis

Judaka

Analysis done in hindsight that utilises logic that wouldn't have been predictively useful is worthless.

Once we know the result, we can emphasise, interpret and characterise in order to lead analysis towards the known conclusion. The logic might appear solid, and the outcome may seem foregone. It will likely appear as though there is something to be learned from the explanation because the logic appears to hold so well. However, the logic holds well because analysis done in hindsight has no risk of being proven wrong since the facts are settled. Export the "learned" lesson to a risky format such as prediction, and the worthlessness of the logic will be revealed.

To begin with, you don't even need to explain things in a sensible way. Why did something occur? "God was angry with us, it's his punishment", if the logic resonates with you, then it's believable. Only by exporting it to a predictive format will the uselessness of the explanation be revealed. How "believable" the logic is doesn't make a lick of difference.

Analysis with hindsight only happens when an event warranting analysis occurs, and not whenever something doesn't occur. Something can have a 0.0001% chance of happening, but when it does the hindsight analysts swoop in and explain how everyone should've known to expect it.

Every news article has the same comments throughout, the wise public comes to offer their "moral" of the story, or their wise advice based entirely on the news article's outcome and nothing else. Every sporting event, the armchair experts come out and explain how obvious it was that the losing team's tactics were wrong, and then say whatever crap they like.

We're all reacting to the recent thing, to the new event. Public discourse is usually orientated around hindsight. this is how people are being taught to think.

These so-called critical thinkers are at the mercy of what they're exposed to, they'll conclude in favour of whatever was the most recent outcome, based on whatever material comes their way. No need to convince with arguments, just show occurrences that conclude the way you want. The public will naturally process everything the way you wanted them to anyway.

jgill

These so-called critical thinkers are at the mercy of what they're exposed to, they'll conclude in favour of whatever was the most recent outcome, based on whatever material comes their way. No need to convince with arguments, just show occurrences that conclude the way you want. The public will naturally process everything the way you wanted them to anyway — Judaka

The antithesis of academic historians' efforts. You're speaking of the babble that comes off the TV or social sites. If so, what's your point? The obvious?

T Clark

These so-called critical thinkers are at the mercy of what they're exposed to, they'll conclude in favour of whatever was the most recent outcome, based on whatever material comes their way. No need to convince with arguments, just show occurrences that conclude the way you want. The public will naturally process everything the way you wanted them to anyway. — Judaka

When something goes wrong, you get new information that might have affected the decisions you made if you'd known it before. Say an engineer designs a building with reinforced concrete floor beams. A couple of years later, after the building is complete and has been operating for a while, cracks are found in some of the beams. Examining the beams indicates that they are failing because bending stresses exceeded the allowable strength of the concrete or the amount of reinforcing steel. Based on observations you make hypotheses about what the failure mechanisms were. You rerun the calculations to see if the beams were designed correctly. You take samples of the steel and concrete to see if they met specifications. Based on that information, you make a determination about what went wrong.

Agent Smith

Well, to be fair to all the hindsight experts, foresight is, I, think, orders of magnitude more difficult. That said, part of the 20/20 vision of the former is information on some folks who did correctly predict, or at least mentioned relevant factors regarding, an event in question, but were ignored for various reasons.

In a sense then those who have the gift of foresight go unheard. This is part and parcel of complexity as far as I can tell and is likely to stay this way until someone who has the goods finds a solution. Until such a time, expect hindsight analyses to dominate our lives as they have and as they do.

Also there's a psychological need to seek explanations for stuff that went right and stuff that went wrong. Perhaps hindsight analysis is exactly what a shrink would order for your mental well-being.

Judaka

↪T Clark

I agree but those are examples of predictively valuable understandings that are proven with results. Engineering is the worst possible example because you're highly results based, you're working with proven formulas and methods, and you can test your theory both in the maths and in results to ensure its validity. Whereas I'm talking about unproven logic formed during the environment created by hindsight being taken as validated by offering a reasonable, or unreasonable explanation of why something occurred.

↪jgill

Well, it can't be that obvious or it wouldn't be so prevalent, but yeah, I'm sure to many it's obvious.

The antithesis of academic historians' efforts. — jgill

In studying history, because we're precisely not analysing isolated incidents in a vacuum, it should be difficult to fall into the trap my OP describes. In any case, you can still employ methods likely to produce predictive value while studying history, it's the logic that needs to be (or is theoretically reasoned to be) predictively useful. There should be an explanation that would have predictive value and some method of determining whether that value is or isn't there.

Though many experts do abuse hindsight, and it's possible to do it as a historian too, this issue shouldn't impact experts who are good at what they do. Because their main aim is to explain things in a predictively useful way, and they should practice methods likely to produce that value.

I'm not saying you can't analyse using hindsight. I'm saying that hindsight is a riskless format while analysing an occurrence in a vacuum. Without taking the analysis outside of that single incident, it will never be exposed to any threatening elements, since it was done knowing the conclusion. Hindsight offers an opportunity to learn, but the value needs to go beyond predicting the known conclusion, and yes, that's obvious, but examples, where that value is falsely assumed to be there, are omnipresent and it was bugging me so I made a thread about it.

T Clark

I'm talking about unproven logic formed during the environment created by hindsight being taken as validated by offering a reasonable, or unreasonable explanation of why something occurred. — Judaka

I'm ok with this, but I can't really think of examples where it applies. Take your example of Monday morning quarterbacking. Trying to blame someone for a single possible mistake in a game is probably silly and unnecessary. But evaluating performance when trying to decide whether to fire a coach might include incidents like that in an overall judgement.

On the other hand, there was a NE Patriot game this year with one play that had a disastrous result. The sportscasters felt they were qualified to blame Bill Belichick, one of the winningest coaches in NFL history, for the decision. I agree with you on something like that.

frank

↪Judaka

That hinges on knowing the outcome. If I turned history into a video game, you could "backtest" your prediction skills to see if you're using your analytical tools correctly.

Judaka

↪frank

Yes, I completely agree with that.

↪T Clark

it's hard to distinguish in your examples whether you're separating the analysis from the result.

Going back to your engineering example, after seeing cracks in the beams, there's a redo of the calculations as well as some testing aimed at identifying the problem. That setup requires, ostensibly, the appropriate calculations to be done properly, which will then help identify a solution that will fix the problem - and may actually have to try to fix the problem. The same methods have been used countless times before, and if it doesn't work this time then something must have gone wrong.

However, I don't want that for my example. I want:
1) To know the result beforehand
2) To never have to take my analysis outside of this isolated incident
3) For my analysis to speak for itself without having to produce results of any kind

This results in an entirely riskless scenario, where there's no opportunity for my analysis to fail. My analysis will always result in the conclusion I knew would occur. Essentially, I'm making a prediction in reverse, I'm working backwards from the result, which allows me to input any reasoning I want.

There are some famous examples like stock trading or crypto, business ideas, and the like. A crypto guru may not be able to know if crypto is going to go up or down next week. Nonetheless, will provide you with a comprehensive report on why it happened and how we could've profited from it after it's already happened. Every idea and suggestion will be correct, because, how could it not be?

A criminal goes to jail for 10 years, gets out and reoffends. The public is outraged, why did this guy not go to jail for life? Obviously, a guy who commits a crime like that is going to re-offend!

How could the West not foresee Putin was dangerous!? Look at this highly specific analysis using only points which lead towards the conclusion of Putin being dangerous, it was all there for anyone to see at any time!

Religion, sports, war, politics, business, news, cultural commentary, you name it, it's everywhere.

jgill

Academically, the study of economics is interesting to contemplate in this discussion. The dismal science is always looking backward, trying to determine patterns that predict certain outcomes. An example is something that is apparent now: the inverted yield curve. In the past this has been followed by a recession. Will that happen in the near future?

When I write a computer program and it fails to deliver the result I want, I go back and analyze, step by step to find my mistake. So the outcome is a mistake, clearly, and I move backward in time to see what occurred.

T Clark

Religion, sports, war, politics, business, news, cultural commentary, you name it, it's everywhere. — Judaka

I don't disagree with the points you've made, I just want to make sure a distinction is made between justifiable and unjustifiable uses of what I guess you could call "forensic analysis."

jgill

However, I don't want that for my example. I want:
1) To know the result beforehand
2) To never have to take my analysis outside of this isolated incident
3) For my analysis to speak for itself without having to produce results of any kind — Judaka

1) A hammer hits my foot.
2) Who cares if this happens to others.
3) I determine I knocked it off the table by accident.

Big deal.

frank

How could the West not foresee Putin was dangerous!? Look at this highly specific analysis using only points which lead towards the conclusion of Putin being dangerous, it was all there for anyone to see at any time! — Judaka

Since he had been repeatedly aggressive to his neighbors, I don't think anyone thought of him as benign. Plus, remember that when the US tried to persuade China to ask Putin to back down from invading Ukraine, they publicly scoffed in a way that wasn't particularly graceful.

I don't think this one counts as hindsight analysis.

Judaka

↪jgill

I'm not against looking backwards at all, and I interpret you to be misinterpreting me. Hindsight analysis is perfectly legitimate. Trial and error and fixing your mistakes are wonderful practices. Your examples are of taking hindsight analysis and exporting it into a predictive format, which is fine. The mistake you made will always produce the error it did, that's an example of predictive value. Establishing a trend or pattern has predictive value. In those cases, you could conduct an analysis and get an incorrect or useless answer, and you'd just try again until you got the result you wanted. It's very intuitive, you are consistently checking your analysis for value, so the problem in OP won't impact you.

You can make a very thoughtful, logical and reasonable prediction while analysing using hindsight, and it can be horrifically incorrect. Anyone who regularly conducts trial and error should understand that.

In social media, news, forums, and many other formats, things don't really work that way. The court of public opinion latches onto appealing reasoning which sounds intuitive or reasonable, especially when the recent results support that reasoning (which of course they will). So, that's where it's a problem.

↪frank

It's definitely possible to have predicted how dangerous Putin was as you suggest, and it's valid for you to say it's not a good example. There are many commentators who provide bogus reasoning for knowing in advance about Putin, but you're right, there are enough legitimate ones and there's a lot we can learn from them.

unenlightened

Analysis is always with hindsight; talk requiring foresight is called speculation.

In social media, news, forums, and many other formats, things don't really work that way. The court of public opinion latches onto appealing reasoning which sounds intuitive or reasonable, especially when the recent results support that reasoning (which of course they will). So, that's where it's a problem. — Judaka

With hindsight, I would say that if you look to social media and news for analysis, you should expect to be disappointed. The weather, the economy, and politics are complex systems that are inherently unpredictable because even for a supercomputer information cannot be complete, and minute causes a butterfly's wing or a meme, can have huge impacts on the future. Furthermore, human behaviour is affected by our theories of human behaviour and a predictive theory cannot take account of the effect of its own prediction, because it would enter a vicious analytical circle. Hence psychology and economics and politics are "dismal" as purported sciences. It's as if the fundamental particles of physics are listening to what the physicists say, and and change their behaviour as a result of each new atomic theory.

jgill

In social media, news, forums, and many other formats, things don't really work that way. The court of public opinion latches onto appealing reasoning which sounds intuitive or reasonable, especially when the recent results support that reasoning (which of course they will). So, that's where it's a problem — Judaka

So, all you are interested in doing is assessing whether or not a commentator gives a "correct" analysis of why or how something happened in the public sphere? For example, all the different scenarios leading up to the January 6th incident. If there is no predictive value in this pursuit, why engage in it? Simply for the feeling of smugness it might provide? Or as an academic
exercise?

Or, if you say,"I find it enjoyable to do this", I will completely understand. I do the same with abstract mathematics. :smile:

Judaka

↪unenlightened

How do you separate good analysis from useless analysis without assessing for predictive value?

↪jgill

I'm not necessarily just interested in whether a commentator gives correct analysis, but in what analysis the public embraces. Of course, it's not like it's rare to see this problem on a philosophy forum either, but from what you've said, it seems unlikely to be an issue for you. I don't purposefully expose myself to scenarios where this problem is prevalent anymore than most I'd say, Maybe it just annoys me more than it does others and so I tend to notice it.

Agent Smith

It's doubtful if the adage once bitten twice shy is true. It seems to lie at the heart of the question. Many people have admonished me, saying "you never learn!" There's this concept which someone to whom I'm a bête noir taught me - institutional memory, no such thing except in extreme cases where the memory is reinforced vigorously e.g. holocaust memorials, exhibitions. Even then, as those who were directly affected pass on, the young invariably fail to connect with that which is being recalled and sooner or later, amnesia prevails, and we're ready to repeat the same mistake over and over again.

unenlightened

How do you separate good analysis from useless analysis without assessing for predictive value? — Judaka

You analyse the predictions after the event. A good speculator makes a fortune, but you tell the good from the bad with hindsight.

jgill

Maybe it just annoys me more than it does others and so I tend to notice it. — Judaka

OK, makes sense. It gets to me at times, also. It's an interesting topic.

Judaka

↪unenlightened

That's just a value judgement lol. If you can tell a good speculator from a bad speculator using analysis, that should be predictively valuable, in helping to guess, at the bare minimum which speculators will see future success due to being "good" or "bad". If you can't do that using your analysis methods, then you're proving that you actually can't tell a good speculator from a bad one and that you're just entirely results based. Which is the exact thing that I'm criticising in this thread as being pointless.

unenlightened

Which is the exact thing that I'm criticising in this thread as being pointless. — Judaka

Yes you are. But I am going to continue to judge foresight by whether it conforms to events and not whether it fits my analysis. As far as I'm concerned the only value of foresight is that it should turn out with hindsight to have been correct more often than not.

Judaka

↪unenlightened

What do you mean by "whether it fits my analysis"?

You can judge foresight, or really whatever else, in whatever way you like, but at some point, the value of your efforts needs to help predict outcomes or it's useless. Expecting existing trends and patterns to be predictively valuable is fine, and there are many valid ways to assert something has predictive value.

unenlightened

the value of your efforts needs to help predict outcomes or it's useless. — Judaka

Indeed, and how do you find out whether or not it does? I do it by waiting and seeing how things turn out. That is to say, with hindsight.

Judaka

↪unenlightened

Sure, it's pretty much impossible to analyse without hindsight, the present doesn't last long enough.

Hindsight Analysis

Welcome to The Philosophy Forum!

Categories

More Discussions