Lets say that normal-quality ice cream is 100 points of enjoyable in some arbitrary units. You, as the consumer of the ice cream, do not have perfect precision.
In a single-shot test, you can tell that 100 is better than 80. In a blind test, you will favor 100 over 90 with a high probability, but not 100%. It would take hundreds of blind taste trials, each time asking “which was better” for you to differentiate 100 from 99.
To put this another way, we can think of the ice-cream quality numbers a bit like chess Elo ratings. An ice cream that “beats” (is preferred over) another with 51% probability is better, but it would take a lot of tests to show it.
I will just assume here that everyone has the same tastes, to make the point more clearly. Obviously in real life some people like some things, and others like other things, but that is not what this post is about.
Sawtooth Wave of Compromise and correction
So far I have just argued for the obvious, that noticeable differences in quality are comprised of many un-noticeable differences added together. What consequences does this imply?
One consequence, is a kind of compromise slope. A manufacturer is wondering whether to use a different sweetener in their ice cream (eg. corn syrup vs. cane sugar). The new sweetener is cheaper. The ice cream quality drops from 100 to 98. But the customers can’t tell, and even the manufacturer might not be able to. So the change goes through with nobody noticing.
Months pass. The manufacturer is looking at a new preservative, Q98 drops to Q96. Vanilla is expensive, some kind of beaver-derived alternative is used to replace some of it (Q94),… things continue until we arrive at Q90. Then, a rival ice-cream maker arrives on the scene, selling quality 100 ice cream. Nearly everyone can tell it is better than the Q90 stuff and some customers move to the up-market product.
Essentially, the model is predicting that compromises in quality will happen in many small steps. But upward corrections will happen in discrete jumps. There is little point in improving your product by an increment too small for consumers to notice.
This idea originally came as a way to explain a pattern I observed in my own cooking. I was reminded of it recently by an Astral Codex review of mashed potatoes [1]. “Proper” mashed potatoes with butter are an obviously superior dish to microwaved instant mash with margarine, but over a lifetime someone can move from the former to the latter in steps small enough to barely notice.
The artist at work
An expert (or artist) is able to spot the positive impact of small changes that others might miss, and can chain them together for a total gain that is obvious to even a non-specialist. Specialists will have many skills non-specialists lack, but this is one.
I once had the chance to play with LaTeX settings for a document with someone more skilled than me in making documents look good. They only made 5 or 6 changes to the font/margins etc, and most of those changes (so me) didn’t produce a noticeable improvement. But, a before/after comparison with all the changes was striking, those small changes had really added up. Across several domains I have found that people more experienced will make certain actions, that to me (the less experienced) don’t seem to really add much, if anything. But, in time, those add up.
It can be interesting to put something obviously bad next to something obviously good, and try and work out what the set of (possibly small) differences are that are doing the work [2].
Utilitarian on unobserved margins
Lets say people can’t actually tell the difference (in a one-shot trial) between experiencing 100 utility points of goodness, and only 99 utility points. You are given a moral dilemma, either a million people will get an experience worth 100 utility points each, or a million + 1 people will get 99 utility points each. The first option gets you more utility total, but if we take the second option we get one more person served and nobody else can even tell the difference.
This situation puzzles me. On the one hand, I feel a strong logical compulsion to the first (higher total utility) option. The fact that the difference is unresolvable for each person doesn’t seem that worrying at a glance, because obviously on a continuous scale resolvable differences are made out of many unresolvable differences added together.
On the other hand, how can I say someone enjoys one thing more than another if they can’t even tell the difference? If we were looking at the lengths of strings then one could in fact be longer than another, even if our ruler lacked the precision to see it. But utility is different, we don’t care about the abstract “quality” of the experience, only how much it is enjoyed. Enjoyment happens in the mind, and if the mind can’t tell the difference, then there isn’t one.
Overall, I think I would take the first choice. This fits with part of a larger thing I have where I think that utility doesn’t need to be “noticed” to be real [3].
This is a confused idea at the moment, but it mostly comes from the fact that I sometimes realize that something has been annoying me for ages, but I didn’t previously notice it consciously. The most extreme example of this for me is sometimes I am trying to read from my screen, then notice my vision is fuzzy in the middle, and only then do I notice that I have been experiencing a really painful migraine for over an hour. It seems paradoxical to say I was in pain, and that was bad, even though I didn’t know it, but that seems the best description of the situation I can find. Once I realize I have a headache I can often identify actions that I took in response to the migraine (procrastinating, skipping my afternoon cup of tea, adjusting my screen brightness repeatedly), but when I took those actions I was in pain, but wasn’t yet aware I was in pain. If I am with someone I know well, they often notice I have a headache before I do.
Quality Precision
Ice Cream Elo
Lets say that normal-quality ice cream is 100 points of enjoyable in some arbitrary units. You, as the consumer of the ice cream, do not have perfect precision.
In a single-shot test, you can tell that 100 is better than 80.
In a blind test, you will favor 100 over 90 with a high probability, but not 100%.
It would take hundreds of blind taste trials, each time asking “which was better” for you to differentiate 100 from 99.
To put this another way, we can think of the ice-cream quality numbers a bit like chess Elo ratings. An ice cream that “beats” (is preferred over) another with 51% probability is better, but it would take a lot of tests to show it.
I will just assume here that everyone has the same tastes, to make the point more clearly. Obviously in real life some people like some things, and others like other things, but that is not what this post is about.
Sawtooth Wave of Compromise and correction
So far I have just argued for the obvious, that noticeable differences in quality are comprised of many un-noticeable differences added together. What consequences does this imply?
One consequence, is a kind of compromise slope. A manufacturer is wondering whether to use a different sweetener in their ice cream (eg. corn syrup vs. cane sugar). The new sweetener is cheaper. The ice cream quality drops from 100 to 98. But the customers can’t tell, and even the manufacturer might not be able to. So the change goes through with nobody noticing.
Months pass. The manufacturer is looking at a new preservative, Q98 drops to Q96. Vanilla is expensive, some kind of beaver-derived alternative is used to replace some of it (Q94),… things continue until we arrive at Q90. Then, a rival ice-cream maker arrives on the scene, selling quality 100 ice cream. Nearly everyone can tell it is better than the Q90 stuff and some customers move to the up-market product.
Essentially, the model is predicting that compromises in quality will happen in many small steps. But upward corrections will happen in discrete jumps. There is little point in improving your product by an increment too small for consumers to notice.
This idea originally came as a way to explain a pattern I observed in my own cooking. I was reminded of it recently by an Astral Codex review of mashed potatoes [1]. “Proper” mashed potatoes with butter are an obviously superior dish to microwaved instant mash with margarine, but over a lifetime someone can move from the former to the latter in steps small enough to barely notice.
The artist at work
An expert (or artist) is able to spot the positive impact of small changes that others might miss, and can chain them together for a total gain that is obvious to even a non-specialist. Specialists will have many skills non-specialists lack, but this is one.
I once had the chance to play with LaTeX settings for a document with someone more skilled than me in making documents look good. They only made 5 or 6 changes to the font/margins etc, and most of those changes (so me) didn’t produce a noticeable improvement. But, a before/after comparison with all the changes was striking, those small changes had really added up. Across several domains I have found that people more experienced will make certain actions, that to me (the less experienced) don’t seem to really add much, if anything. But, in time, those add up.
It can be interesting to put something obviously bad next to something obviously good, and try and work out what the set of (possibly small) differences are that are doing the work [2].
Utilitarian on unobserved margins
Lets say people can’t actually tell the difference (in a one-shot trial) between experiencing 100 utility points of goodness, and only 99 utility points. You are given a moral dilemma, either a million people will get an experience worth 100 utility points each, or a million + 1 people will get 99 utility points each. The first option gets you more utility total, but if we take the second option we get one more person served and nobody else can even tell the difference.
This situation puzzles me. On the one hand, I feel a strong logical compulsion to the first (higher total utility) option. The fact that the difference is unresolvable for each person doesn’t seem that worrying at a glance, because obviously on a continuous scale resolvable differences are made out of many unresolvable differences added together.
On the other hand, how can I say someone enjoys one thing more than another if they can’t even tell the difference? If we were looking at the lengths of strings then one could in fact be longer than another, even if our ruler lacked the precision to see it. But utility is different, we don’t care about the abstract “quality” of the experience, only how much it is enjoyed. Enjoyment happens in the mind, and if the mind can’t tell the difference, then there isn’t one.
Overall, I think I would take the first choice. This fits with part of a larger thing I have where I think that utility doesn’t need to be “noticed” to be real [3].
https://www.astralcodexten.com/p/your-review-my-fathers-instant-mashed
Compare: https://www.gettyimages.co.uk/detail/news-photo/1990s-passengers-on-board-a-london-underground-train-at-news-photo/1074411198 , and https://cdn.britannica.com/49/127149-050-0C1C9430/train-subway-station-London-Underground.jpg
This is a confused idea at the moment, but it mostly comes from the fact that I sometimes realize that something has been annoying me for ages, but I didn’t previously notice it consciously. The most extreme example of this for me is sometimes I am trying to read from my screen, then notice my vision is fuzzy in the middle, and only then do I notice that I have been experiencing a really painful migraine for over an hour. It seems paradoxical to say I was in pain, and that was bad, even though I didn’t know it, but that seems the best description of the situation I can find. Once I realize I have a headache I can often identify actions that I took in response to the migraine (procrastinating, skipping my afternoon cup of tea, adjusting my screen brightness repeatedly), but when I took those actions I was in pain, but wasn’t yet aware I was in pain. If I am with someone I know well, they often notice I have a headache before I do.