The Technique Trap
Feedback is King
We reach Goals by doing something till we get Feedback that we have done it well enough.
This feedback may come from others (a coach or a tool like a weighing machine or video game) or it may come from our own mind (“I don’t think I drew that line straight enough.”).
The point is if the feedback tells us that we haven’t done it well enough, we will try to do it better. We will continue in our efforts. And if it says we have done it well enough, we will stop and go do something else.
Note: I’m talking about balancing feedback loops that go one way. The same idea applies when you have the possibility of going either side of the goal.
It all depends on the feedback measure you choose. If you choose a good one, you will work hard till you reach your goal and then stop. If you choose a poor one, you might think you’ve reached your goal even though you haven’t and thus stop taking action. Or, in some pathological cases, you might keep taking action despite having reached the actual goal a long time ago because your feedback tells you that you still suck and you will never be good enough (ok, that might have been a bit too autobiographical).
But the Feedback Measure you choose determines your actual performance.
Moral of the Story: Feedback is King.
Remember, you cannot do something without a Feedback measure. If you were trying to fill a bucket, you would need to know whether the bucket was full so that you could stop pouring.
Oh, I don’t need to look at the water level in the bucket. I can just use the rate of flow of water and the volume to decide when to stop. Ha! Gotcha, SPK. You and your stupid Feedback measures.
Oh, yeah? How do you know when to stop? Let me guess: you use a clock - a feedback measure. Even if you count “one Mississippi, two Mississippi, …”, there’s still a feedback loop going on - your brain checks each Mississippi to see if it equals “sixty” before it shuts the tap off.
Any time you have a goal, you need to have a feedback mechanism to tell you whether you have reached it or not, and whether you should stop taking action. Goal-seeking requires feedback.
An Old Foe: Naive Realism
Our first instinct is to judge everything by how we feel we have done them. This is our default measurement. Unless we have some external measurement that is right in our face, we will persist in using our internal feedback as our guide. This usually works well. Goal: Need to drink water regularly. Feedback: Am I feeling thirsty?
However, we make a fatal mistake - we believe that what we think is true is actually true. Instead of treating our internal feedback as a fallible piece of information created by a fallible brain, we treat it as gospel. This is Naive Realism.
We think we have achieved our goal (because it feels like we have) and thus stop taking action even though we actually haven’t reached our goal. Worst of all, since we don’t even realize that we’re using such a faulty measure, we don’t feel bothered to change our ways. We remain stuck forever.
So, in pursuit of a Goal, we use our internal Feedback blindly. Certain scenarios predictably fool this internal sense into thinking we have achieved our Goal, thereby lulling us into inaction and keeping us at a low level of performance.
This essential pattern foils all our attempts to perform better. I call it the Technique Trap.
Inception
Why do we accept techniques that have zero evidence backing them? There is a deep and detailed answer to this. But, for now, I have just three words: Heuristics and Biases.
Our mind doesn’t evaluate things based on direct empirical evidence. It uses other - possibly misleading - criteria instead of the actual value.
For instance, we use Social Proof to judge the worth of a product (or idea or person). We see how many other people think that product is good and just use that, empirical evidence be damned. We want things other people want. Well, sometimes people want what we don’t need - a brand new smartphone every year. Sometimes nobody wants what we desperately need - fill in the blank.
Similarly, we use Scarcity as a proxy. It’s fine to value something highly when it is naturally scarce, like land or oil or tasty food. But, people can artificially make you think it is very scarce (“Only on NDTV - Exclusive interview with Mr. …”; “Hurry! Offer valid only until stocks last”).
This is the tip of the iceberg. We have hundreds of mental security holes and an entire class of criminals marketers trying to exploit them. These tactics bypass our conscious mind like trojan viruses. Once the inception is done, it’s too late - these are now our ideas, they must be true.
We can barely resist these invisible insurrections. Asking for empirical evidence at each step takes one hell of constant vigilance. A more scientific culture would help - essentially saying for everything: Show me the empirical evidence or GTFO. But we aren’t skilled at being skeptical. So, we pay the price.
A Can-Do Attitude
Why do we think we can execute a technique when we can’t?
We’ve all been in an exam where we thought we knew a concept and then just sat there completely blank when asked to answer the question. Naive Realism got us.
Why do we end up believing such a lie?
The belief we have is:
I can execute Technique X.
How does Naive Realism apply here?
Naive Realism Attack: I think I can execute what I think is Technique X.
First, let’s see how we misinterpret Technique X but think ours is the actual version.
From One Mind to Another
Techniques exist in our brain. They are neural circuits that let us do whatever is needed to get high performance in certain situations. The various resources we use - books, blogs, videos, etc. - are just there to help us build the neural circuits in our mind.
There are lots of variables in the environment, with a lot of possible values, and thus a combinatorial explosion of possible configurations that the environment can be in. We need to learn what to do in each of those situations to get high performance. Teaching someone a technique is about making them develop the neural circuits capable of handling the vast number of possible scenarios. (There is also the speed factor and creativity, but we’ll ignore those for now.)
A person who knows the technique (i.e., can get high performance in this area) tries to convey his rules for what to do in each scenario into a medium - a book, blog, video, you get it. However, the mediums we have are not rich enough to capture the vast information about all the scenarios in the expert’s head. And even if they were, the expert doesn’t have full access to the actual rules he uses. So, we end up with an impoverished and inaccurate representation of the rules in his head.
Now, the learner comes upon the instructions and builds up his own interpretation of the technique. This interpretation is, of course, deeply flawed because the actions he associates with the words in the medium, say, are probably different from the actions intended by the original expert. In any case, the instructions were themselves flawed representations. Most importantly, neural circuits are built by taking action, not by just learning the abstract rules of a domain (loosely speaking). So, we’ve got a deeply flawed interpretation of the technique and it is not even executable.
The way we overcome this limitation and learn any technique at all is by taking action and asking for feedback from a very rich source of information: Reality. We take action in different scenarios, find out whether it worked, and update the technique in our head. In keeping with the vast number of possible scenarios, it may take a large number of empirical tests before we build up a technique in our head that truly delivers.
Now, that is when everything goes well. Naive Realism is hardly content to stand by and watch us do well.
Many a Slip between the Cup and the Lip
It’s all a castle in the air until the final, crucial step of building up and testing our version of the technique on tons of scenarios. It’s entirely smoke and mirrors until then.
Firstly, the “expert” may not actually be one. He may think he has high performance, but actually may not. Next, the instructions he produces are an inaccurate, low-res representation of his actual rules. But, because of Naive Realism, he will believe that he knows the rules he uses in action and that he has depicted them accurately, and thus will not bother to seek empirical evidence.
When the learner appears on the scene, first, he believes that these instructions accurately represent the expert’s procedure, being oblivious to the wealth of information left unsaid; then, he believes he has interpreted them the way they are meant to; finally, he confuses his abstract theoretical procedure with something he can practically execute. If you think the technique is in the instructions, not the neural circuits, then all you need to do is memorize the instructions. Done.
The upshot is he doesn’t feel like taking the most crucial step of the entire rigmarole: building up his neural circuits by taking action in as many scenarios as possible and updating based on Reality’s feedback. He doesn’t have the technique at all, but thinks he does.
This is why we still suck.
If we do inadvertently take action in some scenarios, we will build up part of our Technique. But, the rest of the technique will remain untested and almost certainly inaccurate (because - Naive Realism). Sadly, we will continue to have faith in it and end up dismayed when Reality eventually proves it to be otherwise.
Yes We Can!
We saw how we end up with “techniques” far away from the things we actually need for high performance. It’s one thing to have the wrong version of the technique and think it is the right one. It’s another thing to believe that you can execute it.
I think that before we get any real-world experience applying a technique, we will judge ourselves by whether we feel we can do each step of the instructions.
Obviously, if the step is simple (“Move your left leg forward”), we will believe we can do it. If we have done similar things before, we will believe we can do it. And if we have failed miserably at it in recent memory, we will not believe we can do it. Putting it together, if it is a “yes” for all the components of the plan, then Technique X is Go!
However, with Naive Realism involved, there’s always a twist in the tale. As we saw earlier, we can misinterpret a step in the instruction to be an action other than what the author intended. So, you and I may believe we can execute the same Technique X (say, the Scientific Method), but mean completely different things by it.
Also, we judge a step by whether we feel we can do it. So, low confidence, exuberance, recent successes, poor memory, and other biases can interfere with our judgment. The worst of it is the Cached Thought - an un-updated belief still held in your mind. Cached Thoughts are responsible for the experience of going to do something you could previously always do and finding, inexplicably, that you aren’t able to do it well anymore, like I found out to my chagrin on a motorbike one day.
The good news is that after we get empirical feedback from the Measurement tool (which will presumably score you high only if you follow the original technique), we will come to our senses and realize that we had the wrong idea all along. Failure does that. The bad news is that this problem of misinterpretation and miscalculation occurs at every level of the plan. So, if actions are hierarchically made out of sub-actions, then you can have a false “Yes We Can” at any level and the whole technique will suffer, making you rear-end a car in mid-traffic.
The Technique Trap
Another teeny tiny little detail about techniques: you have to actually apply your technique to get results. Just knowing it does nothing. But, with no empirical measure, we remain blind to this crucial requirement. If it were obvious that we weren’t improving, we would act right away. But, it isn’t, so we don’t.
This goes a long way towards answering the first Entry Question: Why don’t our awesome techniques give us any results?
It’s because we usually don’t have these awesome techniques. They either don’t work but we think they do, or they work but we have a mangled-up version we think is accurate. In both cases, we think we have applied it enough despite getting no results.
This is what I call the Technique Trap.
That leaves the second question: Why do we remain stuck in the Technique Trap with such low performance?
That Feeling of Awesomeness
We don’t try to maximize our performance directly. We use our feelings as a measure of our performance; specifically, our feeling of superiority over others. We maximize it however we can.
This need not be bad for our performance. With a halfway decent measure, it can pay to compare with others - like using rankings in chess, points in games, or number of views on the Web. Competition is a proven technique for driving progress. Our superiority then is decided by our score relative to others.
However, things nose-dive when we don’t have a decent performance measure.
In domains without good empirical measures, I think we compare on the net awesomeness of the “techniques” we “know”.
Let’s call this the Net Benefits Hypothesis. For the “techniques” we think we know, we add up our “benefits” so far and then the potential “benefits” in the future. We do the same for others and the person with most net “benefits” wins. Empirical performance doesn’t enter the picture at any point.
This sounds a bit far-fetched, but it explains why we don’t get results from our techniques and, more importantly, why we persist in our fruitless behaviour.
Gold Rush
Potential benefits can scale up. You can only delude yourself so much about your current performance - it can be done, but there are limits. It’s much more expedient to stack up lots of potential benefits for your techniques and then just convince yourself that this is what really matters.
If we’re mainly after potential benefits, there’s no pressure to test ourselves empirically at any point. It’s all in the future, and we’re free to make up our own fantasies. Without hard tests, the sham “technique” in our head will not get exposed. So, setting up that sham is all we need for the potential benefits. And apparently since instructions = technique, just “understand” the instructions. Done. We have officially “got” the potential benefits!
Now, the game becomes about accumulating as many techniques instructions as you can, pausing only to make yourself feel you understand them. Nothing more, nothing less. It is a virtual gold rush, for fool’s gold no less.
No need to make sure techniques are backed by evidence. No need to interpret the instructions correctly. No need to build up neural circuits by extensive testing. No need to actually apply the techniques a lot. And, no need to measure our progress empirically. We can skip right over the boring parts and just go for gold! Naive Realism is awesome!
Hoarding Fool’s Gold
The Net Benefits hypothesis predicts that, despite having so many “great” techniques, we will have little or no measurable performance gains. Seems right on. It’s all just potential benefits that we were getting high on.
How to get high on potential benefits?
For one, constantly remind yourself of the “techniques” you know when comparing with others.
Look at those dumbasses. They don’t even know technique X, let alone Y and Z. I am better than them.
I know I have done this a lot, especially when feeling insecure. I whip out my list of unique “techniques” and show how ignorant those other guys are. It isn’t about the empirical performance differences, it is about the mere fact that I “know” these techniques and they don’t. What is your experience?
Next, keep looking for newer and more awesome “techniques”, instead of actually applying the techniques you have. If I were trying to maximize actual benefits, I would apply my current techniques as much as I could. Nope. I’m always looking for the next shiny “technique” I can “master”.
Most importantly, hoard your “knowledge” lest someone else should “learn” it and erode your “advantage”.
Think about the most powerful techniques you know. Have you ever been tempted to keep them a secret? I know I have. And they have always turned out to be useless “techniques” in the end. If I only cared about actual benefits, then sharing techniques wouldn’t be such a threat. But, since I compare with others based on potential benefits, sharing feels like it will remove my edge.
A curious corollary is that when you do share your technique, your feeling of achievement will evaporate almost instantly.
This first happened to me with the idea of how Will Power gets depleted with every use. Before, that seemed like the secret to achieving anything I wanted - I could use Will Power on key decisions instead of wasting it on trivial ones; I could strategically boost up my Will Power by drinking glucose drinks; and so on. The best part was that nobody in my circles seemed to know these ideas. Oh man, if Akrasia was the biggest problem we faced, and if Will Power was the key to cracking Akrasia, and if nobody else really knew about this technique, then I was on the path to World Domination! Mwahahahaha!! Epic Power, here I come!
Then I wrote an essay on it.
Now, suddenly, it was just a common piece of knowledge; nothing special, not so powerful anymore. The simple act of sharing my “insanely powerful technique” with others seemed to burst the bubble.
Once I realized what had happened, I decided to unload three of the most “powerful” techniques I knew, to see if they too were only giving delusions of grandeur - Bayes Theorem, Deliberate Practice, and The Motivation Equation. I wrote back-to-back short essays about them to test this hypothesis once and for all.
What happened? The Aura of Awesomeness vanished without a trace. The delirious excitement I felt before was gone. Poof! And these were, without a doubt, the most potent techniques I had.
Others didn’t even have to read these essays. I just had to imagine them reading it. Sense of superiority instantly eroded. Back to the lab again. (That should have been my first clue about Naive Realism and the Technique Trap. Sadly, the game went on for long after that.)
I realized it was the potential benefits as compared to others that was making me giddy. Neither I nor they had gained any actual benefits. That is the strongest evidence I have for the hypothesis that we compare based on actual plus potential benefits, with potential benefits dominating. 1
These techniques were crutches I leaned on whenever I felt threatened by the actual progress of my peers. I didn’t have the actual achievements to match up, so I stacked up these Insane Awesome Techniques and told myself that I was way ahead of the others - they had no clue about these techniques. I was hoarding fool’s gold.
Corollary: No matter how great you think your technique is, share it immediately. You’ll probably discover that the emperor has no clothes.
Corollary: This illusion of awesomeness based on imagined potential benefits will be rampant when you’re “learning” on your own and have no tests and no other people to compare with.
I see this happening a lot in informal domains, mainly those of self-help and “Rationality”, where you learn by yourself without any hard tests and no other people doing the same thing. So, you are free to claim whatever potential benefits you want.
(And yes, “Rationality” is the domain of this website, which casts doubt on all the great “techniques” I’d previously talked about. Live and learn.)
Spiralling Out of Control
We aren’t content with just claiming potential benefits. We jack up these imaginary benefits even more. Why wouldn’t we? If there is nothing to keep us honest, we can make the benefits of Technique X be anything. And since the potential benefits determine our feeling of superiority, we want to boost them by any means possible.
That is where the Affective Death Spiral comes in. It takes place at the junction of the Halo Effect, the Confirmation Bias, and the Affect Heuristic: believing good things about something makes us believe more good things about it; we seek only evidence that backs our beliefs and ignore evidence that doesn’t; finally, we use subjective impressions of goodness or badness (i.e., “affects”) to make judgments.
The Affective Death Spiral starts off with a really good feeling about something, which makes us believe more good things about it, which is backed by the selective positive evidence we collect, which increases the good feeling we get and so on.
This positive feedback cycle of credulity and confirmation is indeed fearsome, and responsible for much error, both in science and in everyday life.
– Eliezer Yudkowsky, The Affective Death Spiral
I suspect we do this with “techniques” and potential benefits. We don’t have empirical evidence that can kill the Death Spiral. So, all it takes is a strong good feeling to kick off the cycle - and we have that in the sweet feeling of superiority. The Affective Death Spiral starts in full earnest.
This is how we get “Epic Awesome Techniques”. This is also how we get screwed.
Corollary: Ordinary techniques cannot compete with these Epic Awesome techniques.
Ordinary techniques have their potential benefits constrained by the empirical evidence. Epic Awesome Techniques do not. Thus, they win all fights hands down. Look at your own life - what “techniques” do you focus on? The slow and steady ones that have been proven to work or the Epic Awesome ones you read on some exciting website somewhere?
This means that these Epic Awesome Techniques will crowd out the techniques that actually work. We could be getting real benefits from the ordinary techniques but, like the bored middle-aged guy looking for some extramarital spice, we lose ourselves in seductive but ultimately self-destructive affairs.
Why we are still stuck
We remain in the Technique Trap by our own wish. We don’t want to get out. It’s filled with Epic Awesome Techniques - why would we leave it?
We aren’t optimizing for performance, we’re optimizing for superiority, which is given by the net potential benefits of our techniques. We stack up these potential benefits one after the other without even bothering to actually master the techniques. We hoard these techniques and protect them from evil empirical tests. To top it off, we go into Affective Death Spirals around them, increasing the imaginary potential benefits (and thus our superiority) even more.
We do the job superbly well, except it’s the wrong job. Naive Realism keeps us blind - it feels so right. Meanwhile, the actual benefits are nowhere to be found. The boring techniques that really work lie rusting in the garage. We end up with a curious mixture of piss-poor performance and sky-high superiority. (My entire life flashes before me as an example… sigh)
Enough of the bad news. How do we rise up from the Technique Trap like Batman in The Dark Knight Rises?
deshi deshi basara basara…
deshi deshi basara basara… (chant)
Answer: If you actually get techniques that have actually given significant benefits, and if you actually implement them in your life, you will actually get those benefits too.
There’s no magic to it. And it’s not a piece of cake. But if you really do it, it will work.
Every now and then, someone asks why the people who call themselves “rationalists” don’t always seem to do all that much better in life …
– Eliezer Yudkowsky, Make an Extraordinary Effort
This is my Fully General Answer to the question “Why don’t Rationalists end up winning like hell?”
Because Naive Realism gets us all.
Notes
Naive Realism doesn’t feel like it’s a sufficient reason to explain the massive failure of so many “Rationalists” to get great results.
To our mind, it feels like there must be some other reason for our failure. Maybe there is some shiny, new technique we have to learn that will fix our problems once and for all. A Silver Bullet that will kill the werewolf of underperformance.
Beware! Great disasters don’t need great causes. The humblest gear in a machine is enough to bring it to its knees.
Several people have questioned why Rationalists aren’t ruling the roost.
Two of the most scathing critiques of the current Art of Rationality are
Ideas generated during the essay
- Cached Thoughts (late at night)
- Naive Realism all the way down
- Technique is in the head
- You don’t have free access to the Technique in your Head
- F*ck! There is no such thing as a Technique!!!
- The thing we pass around in books or videos is just a bunch of words! It’s NOT the Technique.
- Consuming Learning Resources != Building up the Technique
- Complexity (here) = Lots of variables, lots of possible configurations that you have to be ready for.
- It’s the Net Benefits Comparison Hypothesis, not the Superiority Hypothesis.
Summary
Hypothesis: personal achievements <- idea, practice, application.
However, when you don’t have any empirical measurements: respect <- potential benefits of the idea alone!
When you have empirical measurements: respect <- empirical progress, not the potential benefits of the idea or the amount of “practice” or whatever.
Those essays didn’t even go deep. They were just perfunctory descriptions - you couldn’t really do much after reading them - but that still did the trick. My crash wasn’t because I empirically found that the techniques didn’t work; it was because I just shared the names of the techniques with others. That’s all - the names alone were enough to erase my perceived superiority.↩
comments powered by Disqus