Hi, everyone. Welcome to Nontrivial. I'm your host, Sean mcclure. In this episode, we're going to take a look at the notion of quality, how to define it and how to assess it in real world situations. And I'll discuss what I think is a useful heuristic for sizing up both things and people that make their way into our lives. Let's get started. OK. So quality is important in our lives, obviously, right? We want to create and consume quality things.
Uh you know, if we're talking about making things, we want to build quality stuff, you know, we want to build quality software, make quality paintings and sculptures, write quality articles or whatever it is that we do to build whatever we, you know, we we're releasing to the world, we want to be of high quality. And then we also want obviously quality things in the stuff we consume, right? I want to eat quality food, read quality books.
I want to meet quality people, whatever finds its way into our lives, we want to be high quality and that's going to overall just kind of improve the quality of our lives, presumably, right? So we all have this notion of what quality is and why it's important. But when it comes to consuming things, we actually don't have much control over it. Right.
So again, going back to the stuff we build, you know, sculpting, painting, writing articles, giving presentations, you know, forming companies, you know, giving speeches, whatever it is, writing software, anything that we build, uh we have the ability to, to kind of fold the learnings in, we have the ability to iterate and rearrange, you know, it's an open system, right? That we can continue to improve over time.
In fact, in the things that we create, you can almost equate quantity to quality, right? At least there's a kind of equivalence there. And that's because the more we do, the better we get, right? So we have that control. But when it comes to, you know, buying things or, or just consuming things, reading things, even meeting people, we don't have any say in the quality, right?
Whatever the quality is uh that is released to us that, you know, based on something that we purchase or something, you know, someone that we meet, that's what we get. And so the assessment of quality ends up being pretty important in our lives, right? Because we don't, you know, again, for the things that we consume, we don't have control over it. So we have to be able to assess it.
Uh you know, I say purchase time, you know, or maybe you're doing job interviews, you need to bring someone new into your company, you need to be able to assess the quality of the individual, so to speak, right? So to really get at how quality and decision making overlap, we need to land on a definition of quality that we can use, right? And this isn't necessarily as easy as we might think.
Again, we all have the kind of an intuitive sense of what quality means, but to actually have a utility, do it to make it useful to how to go to how to go about assessing the quality of something, you know, before you go ahead and make that purchase or go ahead and hire that person or whatever it is. Um how it actually overlaps with decision making is a little more nontrivial, you might say, and we can't really do this in dualistic terms, right?
It's not super useful to be, well, there's high quality and there's low quality. You know, if I'm gonna go to the, I gotta go buy a coffee table and I gotta go to the local craftsman who does it, you know, he, he's, he's, he's just kind of his own little business and he, he sources all the quality materials, you know, that's obviously higher quality. And then if I go to a global furniture company, uh you know, that coffee table is gonna be much lower quality.
So I know there's one that's high and one that's low. I'm just gonna go for the high. Well, that's not super useful because everyone has a requisite level of quality that they're willing to pay for. Right. Um Yeah, obviously the, the kind of local craftsman who makes his qua coffee table is going to be of higher quality, but he's also going to charge more for it as he should. Right. Uh But that's not necessarily what I need.
If I'm moving around all the time, maybe I only need a table to last, you know, two or three years, then I don't mind replacing it, especially if I didn't have to pay that much for the table to begin with. Right. So everyone has kind of a requisite level of quality that they're willing to pay for.
And so what that means when it comes to quality, whether we're talking about coffee tables or food that we eat or people that we hire or people that we meet, whatever, there's some requisite level, there's a gray area, there's that kind of complex, non-trivial middle ground that we operate in, in real life. And that's where you have to kind of strike the balance, right?
How do you make a quality assessment about the things that find their way into our lives in such a way that you can kind of strike that right? Requisite level, you know, not too much, not too high, not to do it in dualistic terms as just high quality, low quality, you know, how, how do you, how do you strike that level where the quality is acceptable? So I think that's, uh you know, a really important heuristic to come up with in life.
I think it's worth stepping back and thinking about how do we assess the quality because it comes up in all areas? I mean, I've already given some examples and food is maybe one of the best examples. We're always trying to, always trying to think about what is healthy, right? Um You know, do I obviously I want to eat quality food and maybe I've got a sense of what is quality and by quality, I I typically mean healthy, right?
I mean, quality food is going to impart that nutritional value to our bodies the way it's supposed to, it shouldn't be doing other things, right? It should be just doing that and also giving us, you know, the flavor and texture that we expect and beyond that, you know, just leave it alone, right? I just want that high quality thing.
Um and we all kind of have maybe a, a kind of intuitive sense that less ingredients are more or maybe that um something that's locally made is a, is a better quality and often that's true. But we don't necessarily have a robust definition of why that is and there are cases where that is not necessarily true. Someone might be using less ingredients but they're poor ingredients or they're just not higher quality, someone might be doing it locally, but they're still using crap stuff, right?
You need a more kind of robust way to defend the heuristics that you use in life to assess quality. OK. So quality is really hard to define, let's think about that for a while. Why that is and then we'll kind of move into some of the mechanisms and solutions as I usually do. So when we say something is high quality, we tend to mean that it's not gonna break, right? It's gonna go a long time without breaking basically.
OK. So the coffee table example is a coffee table that is going to continue to be a coffee table, it's not gonna have one of its legs break just because I put a few heavy out, you know, things on top of it or I shift it around to different parts of the room. It's it's uh it's going to not break and this is not just for physical things, right? You can think of this more appropriately on informational terms, you know, as I often argue, uh you non-trivial to think of things information, right?
So if you think about eating something, you know, at first it's like, well, things break when you eat it. So what do you mean? Right. But information, it doesn't deviate from its intended purpose, right? The the food is meant to impart a certain flavor and texture and a certain nutritional value to you. And you know, for the, for the time period in which you are consuming that food, it's not supposed to break from that. Right. It shouldn't be doing a bunch of other stuff.
Um, a book that someone writes, if I'm reading that book, that book is going to be higher quality. If it doesn't deviate from the main kind of theme or plot, you know, or, or, or narrative structure to which it's supposed to be accountable to, right? If it goes off with too much rhetorical flourish or even being too pedantic and getting too detailed, it's breaking from away from its main purpose. It's deviating from that coherent message that I expected, right?
A high quality piece of writing isn't gonna deviate that much, right? Um It's gonna be more coherent. Uh you could really use this for anything. So, you know, this food writing, you know, people even, right? OK. So what do you mean like people not breaking well, I mean, if I meet someone who's high quality, the longer they go uh staying, you know, without deviating from, let's say their core set of values, which is a set of values that were originally aligned to mine.
And so therefore, I expect those to stay the same as they do for me as the longer they go without deviating from those core set of values, then you could argue that, you know, that that's a higher quality individual, right? You know, and obviously, I don't mean, like intrinsically they are higher quality. I mean, everybody's the same that way. I just mean quality in terms of how, well, how commensurate it is with your set of values.
Let's say if I'm hiring something into my com someone into my company and they give themselves a title, you know, they are, you know, a software engineer, then I expect them to not deviate from the skill set of a software engineer too much. If they do, then maybe we place them somewhere else and maybe they need a different job.
But in terms of being a quality candidate or individual for that job, they shouldn't break from, you know, that that original purpose that they had those outputs that I expect them to produce. Right? So no matter what it is, you know, coffee tables, people food, any time you come to assessing quality, we generally think that it's not going to break over a long period of time. That's what quality means, right?
And the deeper notion at play here and the one that's probably a little more useful is uh you know, quality is, is the ability of an object to solve the problem for which it was intended.
So when something deviates, you know, like that candidate deviates from software development or that book deviates from the main, you know, plot of the book or narrative structure, you know, it's, it's getting away from its intended purpose, it came in, you know, with the title of the book or the title of the chapter, it was supposed to be talking about this.
And if it's quality, it's going to maintain that purpose, something that deviates from that intended purpose gets away from that, that food is expected to impart the outputs of flavor, texture, and health. And if it deviates from that, it's of lower quality, right? So one way to capture the notion of an object solving its appropriate problem, which is what we're talking about here is through category membership. And I just mean, you know, we go through life and we form categories all the time.
We have to, that's how we make sense of the world. I call things tables and dogs and cats, you know, we we create these abstractions because there's too many details to handle. So we have to raise the level of abstraction in order to anchor on to something that we can understand, right? That's how we comprehend the world. We give things labels, we give things categories.
And so if something belongs to a category, it's because you know all the instances of those objects kind of share some characteristics at some level of abstraction, right? All tables uh fit into the category of tables because you know they have four legs and they yada, yada, whatever it is. Well that yada yada is really the problem that it's meant to solve, right?
The the table is a table because it provides a flat service with which somebody can place objects on top of and which someone can work on or something like that. Right? It has a definition to its purpose. It's meant to solve a specific problem. That's why a table is in the category table. If the table breaks its leg, it doesn't belong to the category table anymore because it's not solving the problem. A table is meant to solve anymore, right?
But category membership doesn't really help in decision making by itself. Because if you think about it at purchase time, let's say both a high quality and low quality version of a thing belong to the same category, right? At purchase time, a knockoff purse is a purse just as much as a blue crocodile Hermes, Birkin handbag, right? They both belong to the category purse and yet they differ greatly in uh in uh in quality, right?
The local craftsman's coffee table is a table just like the global furniture company super, you know, let's say, you know, particle board, cheap glue crap table, that's still a table. They differ greatly in quality, but they both belong to the category table. OK. So quick recap. I said something is a high quality if it goes a long time without breaking. And if you think about that on informational terms, I think this would apply to everything.
The deeper notion at play here is that an object that solves the problem for which it was intended can be thought of as high quality. Because if you're solving the problem for which you were intended, you're not breaking away from that. And you could frame that in terms of category membership. As long as that object keeps solving the problem for which it was intended for which it was made for, then it belongs to that category.
So as long as it sticks in that category, we can think of that as high quality, right? Or the longer it goes sticking into its category, right, the higher the quality is and and there's gonna be re you might think that's kind of a funny way to define quality. Like why are you talking about category membership? I'll explain that in a bit because it relates to the kind of more physical real world heuristic that I want to kind of come up with to assess quality, right?
But the category membership by itself isn't really that useful because again, low quality stuff and high quality stuff both still belong to the same category, right? The knockoff purse and the Birkin handbag are both a purse handbag, right? Um Like the coffee table from the local, the local shop versus the global, you know, they're both tables. So category membership doesn't really get at it quite yet.
What makes assessing quality challenging is that at decision time, we don't know when something will bounce out of its category. Yeah. Right. And, and yet that's the only piece of information that really matters. So we need a more robust definition of quality, something that's less kind of hand wavy about, you know, things not breaking and, and things deviating from the problem for which it was intended for which it was made for, uh, you know, category membership. These are all true.
I think they're conceptually true. They make sense. I think they've got, you know, uh, they, they're, they're fairly intuitive. I think most of you would agree that, you know, when I talk about those things, yeah, I think that makes sense, but it's still kind of hand wavy and it's not super useful because again, at the surface, I don't know if something is going to bounce out of its category or not, right? At purchase time, at decision making time, right?
When we go to assess the quality, how am I supposed to know if something is going to bounce out of its category or not? We need a more robust definition of quality, I think to help with that assessment and to be robust, it's got to connect to something that we can, you know, assess in real life. And that means it's gotta be something physical, we can latch on to. It's gotta be something real.
So beyond the concepts of breaking and problem solving and category membership, you know, what is it I'm looking at in the real world that connects back to those things. Ok. So this leads me to what I believe is really the only definition of quality that is robust to some of these examples that I'm given these examples that I've given a high quality object is one that is more likely to maintain its category membership over time. OK. That's my definition.
A high quality object is one that is more likely to maintain its category membership over the time. It doesn't sound that different than what I already said. I already talked about category membership. But notice I'm using the word likely that adds the notion of probability and, and that makes it a much more practical definition because at purchase time, right, you don't have much information.
But if you can somehow assess the likelihood of something bouncing out of its category, you're not saying whether it did or not bounce out of the category because again, at the beginning, knockoff purse, crappy p you know, the crappy purse on the good purse, the, the crappy table and the good table. They both have the same category, right? The low quality coffee and the high quality coffee are both coffee, right?
So that's not really useful, but they should have a difference in the likelihood of bouncing out of that category. And if we can assess that, that would be useful, that would be a practical heuristic.
And that would fold back to all these other definitions of category membership problem solving and you know, the the deviation or the breaking from, from, from the intended purpose, you know, all those concepts that seem to make sense with the quality, that would be kind of like the physical real thing. If we could latch on to something that allows us to assess the likelihood of something bouncing out of the category, even though at decision time, they both belong to the same category.
If we can do that, that would be much more practical. So hopefully, that makes sense. OK. So I would argue that that's only possible if there was a notion of degradation of the mapping between solution and problem. So again, I said we've got these definitions of, you know, something belongs to the category if it's solving the problem for which it was intended, right? So there's this this kind of connection or mapping between solution and problem, right?
A table is a solution that maps to the problem, provide a flat surface for which people can work on. The cracker is a solution to the problem of provide a thin wafer that allows someone to consume cereal grain or whatever, right? Everything that is created is a solution to a problem. There is a connection. That's what I mean by mapping, mapping just you know some solution to problem. There has to be a notion that even though two things belong to the same category, right?
The crappy coffee and the good coffee, the crappy table and the good table, they both copies both tables. One must have a degraded mapping between it. It's, it's still solving the problem that it was made for. But it's like a weaker link. And if you could somehow assess that weaker link between solution and problem that would allow you to say this one has a higher likelihood of breaking away of, of losing its category membership at an earlier time compared to the high quality thing. Right?
Think of it just a string, right? There's a string connecting a solution to a problem. Imagine a string with four strands versus a string with three strands. The string made of four strands is is a tougher string and the string made with three strands is a weaker string. So I'd expect the three strand string to break earlier. OK. That's conceptually the same idea if we could somehow get at the strands between the solution and problem.
If I can look at a particular coffee table and again, it's probably in the middle ground. It's not necessarily just comparing the local craftsmen to the global furniture company. I have some requisite level of quality that I want to be able to pay for uh the quality candidate that comes into my company. You know, I have some requisite level of quality, right? We got to be realistic. It's in that gray nontrivial area.
But if I can somehow assess the number of strands right, in between that solution and the problem that person or thing is supposed to solve, then I can get a sense of the likelihood that it will break either earlier or later. OK. So it's this, it's this notion of what I call degraded mapping like in the mapping, I just mean the connection between the solution and problem degraded would be like less strands in the string, right? As though it was kind of like rusting away a little bit, right?
A solution with a degraded mapping still belongs to its category, still a software developer, still a cup of coffee, still a table, still a book. But it has an increased likelihood of bouncing out of that category by virtue of a weaker link between that solution and the problem, it was made for a knockoff purse should be considered to have a degraded mapping between itself and its problem which is provide a bag to carry, right, put stuff in despite still being called a purse. OK?
So just, just really quick recap because I want to make sure I'm clear here, right? We understand what quality is. We don't want to deviate. We don't want to break. I related that to category membership. But I said that, you know, look low quality and high quality things are both in the same category.
So the only way that's useful is if if if you could assess the likelihood of something bouncing out of its category, right, if you have a higher likelihood of bouncing out of the category, it's probably going to break earlier, right? And if, if, if you have a, a lower likelihood of bouncing out of the category, your category membership is stronger, you have a stronger link between the solution and problem and you're probably going to last longer.
And if you could look at all the things that we can at any given time in life, if you could assess what you're consuming and say, I think I know how many links are in that strand or sorry, how many strands are in that link? Right? II I have an idea, a real physical way with which to assess the likelihood of something bouncing out of its category. Then I can, I can I can come up with a much more real world heuristic to assess the quality in that middle ground. OK. I hope that makes sense.
So the big question now of course, is, well, what causes an object to have a degraded mapping between the solution and problem? I like, what is that? Right? What would make something have less strands in the string that connect the solution to the problem? What would weaken that even though it belongs to that same category, even though it's still solving that problem, what would make that link weaker? It's not something that can be detected at the surface.
Since again, both high quality, low quality things are members of the same category. Purchase time, right? It's gotta be something internal. OK. So what do I mean by internal? I mean the the internal ingredients of something, the guts of something, right? Not something that's immediately apparent uh on the surface, not something we perceive right away something about its internal makeup. OK. So let's let's explain how we connect this to this.
So consider the mass production of the coffee table, right? A mass produced table, it doesn't just solve the problem of providing a flat surface to work on. It actually solves a number of other problems related to mass production, right, a mass produced table must be easily transported. It means it must be easily packed into tight spaces, you know, be relatively lightweight and easily reassembled.
A mass produced cracker isn't just solving the problem of providing a crisp wafer for the convenient consumption of cereal grain. It's also solving the problem of having a long shelf life remaining intact during transportation, you know, having a consistency in flavor. When we mass produce something, we are forced to solve additional problems related to mass production in this example.
And it's those solving of additional problems that causes the degraded mapping between the original link in solution to problem, right? Originally. So, so if we go back to that local craftsman and say, you know, so he's building high quality tables. He's been doing it for years, he gets really popular and now he wants to scale and now he needs to produce millions of these tables.
Well, that's not possible by just taking the table that he's making today and say now we'll just go make millions of it. That's not going to work. That type of construction isn't going to be economic goal, right? It is not going to be possible that way, it's not going to fit the assembly line, it's not going to yada, yada, all these extra concessions are going to have to be made in order to fit that table to mass production. You can't just mass produce the same thing, right?
You got to make these concessions and that is the cost of doing something like mass production. You can still do the table. You might even be able to make it look pretty much the same, you can give it the same name, the the creator's name could be, it's still a table still belongs to the category, but there are internal guts that are changing in that table in order to solve the additional problems that go along with mass production.
OK. Going back uh you know, towards the beginning, I said that, you know, when we create things ourselves, this is like an open system where we have uh information kind of pouring in and, and, and in open systems. One of the hallmark, you know, kind of properties of open systems is that they can lower their entropy, right?
Lower their kind of uh messiness because we can use feedback and information and contemplations to continue to tweak and, and uh and rearrange and Polish and edit what we do to, to make it more cohesive, to make it cleaner, you know, to kind of re factor it right. Open systems can lower their entropy over time. But mass production with this example is a closed system because you're just replicating, right?
It's a closed system and closed systems either see their entropy stay the same or much more likely see their entropy increase. OK. And so replication is even though it's still a table that's getting produced internally, it has to make these concessions in order for to, to make mass production possible. And that's increasing the internal guts, the the entropy of the internal guts of the thing. OK. So, so a mass produced table isn't gonna have a nice lo locally sourced solid oak wood.
It's going to have, you know, particle board and maybe cheaper resins and kind of wood veneer or, and maybe, you know, you can go look online, you can see some of these kind of corrugated type cardboard pieces that they use inside the the the thinner layers of particle board, right? It'll look like a table, it'll function like a table. You know, if it looks like a duck and quacks like a duck doesn't mean it's a duck, right? Internally.
Well, well, it is actually it is still the duck, it is still the object. It is still the category. But internally it's got all this difference compared to the local craftsmen. OK? So again, I said you know what causes an object to have a degraded mapping between solution problem is not something you can really detect at the surface. Because if you just look at a table, it looks like a table. If you look at a piece of a cup of coffee, it looks like a cup of coffee, right?
If a candidate first walks into my room, whatever it is, right? Um There's nothing in there that's going to tell you the degradation unless you can somehow get at something more internal. And and the reason why the internal guts are changing between low and high quality is is if you go to do the low quality thing, you're gonna have to make concessions that change the internal guts of something.
And that's what causes even though you still belong to that category, there's a a weaker link between its ability to solve the problem. It's a type of degradation because there's a cost to it. OK. So to be clear, if we're looking at the low quality table, yes, it does still solve the problem of providing a flat surface just like the local craftsman's table does. But it's also made of particle board and, and car corrugated cardboard pieces and lower quality probably resins and whatever else.
OK. So there is a degraded mapping because now it's more likely to pop out of the category that we call table than the local craftsman table is, right? It it's more likely to lose its category membership and we'll do that at an earlier time. Now you might say, yeah, that's obvious. I know a global, you know, furniture company has low quality crap and I know that the local person, but again, those dualistic terms are not useful. We're in the gray area, we have a requisite level.
So the, so, so what we're trying to, to get at is the ability to assess the internal guts of something, the ingredients of something. So the question is, how do you do that? And we probably all maybe think we already know how to do this. Right? So food again is such a good example, right? If I am going to purchase a piece of food and yes, I've got some middle ground gray area, requisite level, right? But I still know how to assess that.
Don't I like if I'm in the airport, I'm not gonna eat the healthiest thing. I'm in a rush, I'm starving. I'm probably gonna eat a mass produced snack once in a while in the airport. I probably have to. But if I'm at home, I'm not gonna stock my shelves full of that crap. Right. I'm gonna want the healthy stuff and I think I know how to do it.
We, we, you know, we turn over the labels, we look at the ingredients and we make the assessment, but maybe it's not always that easy because what is the rule there? Is it just less ingredients or better. A lot of times that is the case if I have fewer ingredients, uh it could mean that it's higher quality if you go, if you use that coffee table example, right? The local craftsmen, his coffee tables probably have fewer ingredients.
It's probably a hunk of oak or some wood, some glue and nails, right? I mean, there's not much more, I mean, maybe he's putting some varnish on top or something. But really if you look at the ingredient list of a locally produced high quality table, probably less ingredients than the global furniture company's table which has particle, you know, it's got like wood chips and resins and epoxies and then wood veneers and then cardboard and whatever else.
Because again, they're making all those concessions for mass production to get it to do all those additional problems that you know, degrades the the solution to problem mapping. It typically does have to have more ingredients. So they say, OK, so there we go, we've got a rigorous definition. Let's just have fewer ingredients, but that's not necessarily always the case. Let's use an example of shortbread cookies. Maybe somebody, somebody's grandmother makes shortbread cookies.
They've done it all the time. It's only three ingredients, flour, sugar, butter. OK? Now let's compare this to something that's mass produced at the supermarket that brands itself as healthy keto cookies. And it has four ingredients still pretty low. But more those ingredients are almond, flour, maple syrup, vanilla and salt. So now we're comparing the three ingredient, shortbread, cookie of flour, sugar, butter to the four ingredient cookie of almond, flour, maple syrup, vanilla, salt.
Well, which one's healthier again, kind of equating quality to health here. And I'm not, you know, this is beside the point of, you know, somebody might not be making shortbread cookies for the sake of health, right, or whatever it is. But as a way of assessing the quality of what we consume, if I really want to eat a cookie, and I want the higher quality one which means yes, flavor, yes texture, but also imparting, you know, some level of nutritional value.
Are you gonna go for the flour, sugar, butter one or the almond flour, maple syrup, vanilla salt. Now there might still be some debate here. But what this example shows is that there are scenarios where more ingredients could actually be higher quality, they could be healthier. Even the coffee table that gets made with more ingredients could be better than a fewer ingredient, coffee table.
If a fewer ingredient, coffee table was using a bunch of uh or, or actually just a few lower quality ingredients itself, right? Maybe they didn't source the the the the the individual wood pieces locally, maybe they cut corners on the uh on the, on the resin or the glue that they're using whatever it is. OK. So there has to be something a little bit deeper. A little bit more profound when it comes to the heuristic than just fewer ingredients. OK?
Even though that might cover quite a few cases, there are, it's, it's not the most fundamental thing at play here because there are cases where you could have more better ingredients than fewer worse ingredients, right? The quality of something could be better with more ingredients. So, so what is the more fundamental thing at play here? And what it comes down to is it is it it it gets back to this notion of entropy, right?
Remember we said, lower quality things tend to have to make concessions and and the way that plays out is that the internal guts end up being higher in entropy, they end up being more heterogeneous, right? They need that heterogeneity because they need to solve all these additional problems beyond just the original solution to problem mapping beyond just provide the flat surface, it's also got to be reassembled. It's also got to fit the assembly line.
It's also got to be lightweight and transported to do all those other things. It needs to to to to basically take on different configurations. And the way it does that is by being more heterogeneous, but more heterogeneous does not necessarily mean more things. So let's do a little exercise and and see if you can follow. So grandmother's cookie is flour, sugar and butter. Now, what are those ingredients doing? They have their own kind of internal purpose, right.
Well, flour is like the material to kind of just make the bulk of the thing. Sugar is for flavor. And you could say butter is for texture, right? And we're simplifying things a bit, you know, maybe butter plays different roles. But, but basically we could say what's happening in a three ingredient, shortbread, cookie made of flour, sugar and butter is we are taking care of material flavor and texture. Ok. So kind of three dimensions.
If you will right now, let's go to the mass produced larger ingredient list of almond, flour, maple syrup, vanilla salt. The almond flour is also material. The maple syrup, vanilla and salt, you could argue are all for flavor, which means that the mass produced cookie is really only doing two things. It's providing material and providing flavor because the maple syrup vanilla salt could be combined into a single group. II I, it's operating on the same dimension.
It's really just there for flavor. And so the mass produced cookie by only doing material and flavor, even though it's got four ingredients is actually less heterogeneous. Whereas the grandmother's cookie of flour sugar butter is operating across three dimensions. It's got material, it's got flavor. It's also imparting texture is actually more heterogeneous. Now, here's the point to relate it back to category membership. Let's think about how we make a category for something.
There are a bunch of different instances of things and then we group them together and we give it a label, right? So if we're talking about dogs, we take a look at all these different breeds of dogs and we say they are all a dog. Now, how are we able to do that? Well, it's because all those different breeds of dogs are sharing a bunch of characteristics, right? And they're sharing them in ways that other objects don't share them. Right? The cats don't fit into that category.
So if we're taking a look at uh a bunch of different breeds of dogs, well, they have four legs. Uh now that's not enough because so do cats. Well, they have uh let's say four legs and they always have fur and they have teeth and on and on and on and we, we keep going across a number of different dimensions and we say that they seem to kind of take the same values across those dimensions, right?
And if we were to take, you know, those dimensions and apply them to cats, they wouldn't have the same values or they wouldn't even have those same dimensions, right? We group things together across a number of dimensions and then we apply a category to it. Well, the more cohesive, those uh dimensions are the easier it is to define the category. So, you know, going back to the fur example, on the dog, there are some breeds of dogs that say that don't have fur.
So maybe that's not actually a dimension we can use or, or there's a bunch of different values that are kind of fur. It's kind of not fur. Maybe it starts to get really heterogeneous. There's a lot of dimensions that go into trying to say how these different animals are related in a way that we can call them dog. Ok. It's not just, you know, the fur, it's not just the ears, it's not just the teeth, it's not just the four legs, it's not just the tail, like a lot of animals have those things.
You really have to kind of extend out or increase the number of dimensions that you by, by which you compare these things that we call dogs in order to create the category dog. And the more the dimensions you have to use, the harder it is to really define the category, you know, is, is a dog that doesn't have hair. Is that still a dog? Well, yes. OK. So what are those dimensions? Right?
You have to use more of them whereas something that is more cohesive and has fewer dimensions by which you compare the different instances. Those things belong to a category in a way that is much easier to define the category. If I play something into the category pencil, this is a lot easier to do. There's not as much variation of variety like we see in, in the the dog category, right? There's not all kinds of different, I mean, there's still a number of dimensions.
But, you know, you've, you've got graphite in the middle and you've got kind of a wood exterior and you can, whatever, whatever the number of dimensions I would argue is much more cohesive. It's easier to define the pencil category than it is the dog category. You're probably gonna find a lot more examples of dogs that are kind of fit into the category. But oh, now it's a bit of a stretch than it is for something like a pencil. OK. And maybe these are not the best examples.
But the point is is that the more cohesive the members are of a group, the easier it is to give that group a label. And I think that's pretty intuitive. The more heterogeneous, the characteristics are the more different ways that members of a group seem to group together. The more different ways there are, the harder it is to give that a category to give that a label. And so that means that the more cohesive, the members of a group are the more robust that category is over time.
So let's go back to the four ingredient cookie. We said that was actually less heterogeneous because it's really only operating across two dimensions, material and flavor. So that's a very cohesive description of a cookie.
And that means that I would expect that cookie to maintain its category that we call cookie or shortbread cookie over time, more than something like the three ingredient, which was more heterogeneous, the more heterogeneous cookie is still called a cookie, but it has more dimensions by which it's operating material, flavor and texture. And so it could be expected to actually bounce out of its category at an earlier time.
Now, remember we're only interested in the likelihood of something bouncing out of its category. It doesn't matter, you know, we, we're not saying, you know, grandma Sharri cookie is gonna not be a cookie. By the time we go to eat it, that's not what we're saying. We're saying that these, these two different versions of cookies are both cookies, they will both be cookies when we eat them.
But one has something about its internal guts that give it a higher likelihood of bouncing out of the category. It's not that it's gonna bounce out of the category before we eat it, but because it has a higher likelihood of bouncing out of its category, it's got something in its internal guts that degrades the mapping between solution and problem that makes it solve problems beyond just what it needs to do, which is just to be a cookie.
And because of that internal description, that heterogeneity of that internal description, it's actually a lower quality thing to consume your body only needs the cookie to do, you know one or very few things, right? It only has to, you know, taste a certain way and not do something to my body that I don't want it to do, let's say, right.
Uh And, and the point is, is that the more cohesive, the description of the object, the more likely it's only going to do the thing for which you purchased it, the, the, the purpose for which it was intended, right? I'm only buying this cookie or consuming this cookie to do what a cookie is supposed to do. That's it. And I might not know the other things that that cookie might be doing right.
But the more heterogeneous its description, the more likely that it is going to be doing other things that might be other health things that you don't want it to happen, want to happen. It doesn't really matter. The point is that it's going beyond just its intended purpose. It has a higher likelihood of breaking from that and maybe that break is happening, you know, in inside the body or whatever it is. Let's go to the back, you know, back to the coffee table.
Um you know that coffee table just needs to be a coffee table. It just needs to provide the flat surface as I have a nice direct mapping between solution and problem. OK. That's what it was made for. But if you go to mass replicate something mass produce it, you make those concessions, you increase the heterogeneity of the internal description of something that's being done because you got to make all these concessions for mass production it's got to solve all these other problems.
But that doesn't, that's not something I care about. I don't need it to solve all these other problems. Right. I don't need all that mass production. Right? I just want the table. But if something is solving a bunch of additional problems that's going to degrade the mapping between the original solution and problem, there's always a cost to doing that. OK. So let's just to wrap this up and I'll use some, some final examples at the end here.
So really it comes down to how well we can compress the ingredients of something. So if you're gonna go into the store and you're gonna turn that wrap around and look at the ingredients just to take a look at the ingredients. And don't just think the number of ingredients. Although that is kind of a first step like if something's got 60 ingredients, you're probably not gonna be able to compress that that much anyway, right?
So that, that there's probably just a lot of dimensions there anyway, it is really heterogeneous. It's, it's not gonna be good for you. But if it comes down to, you know, again, trying to get that middle ground, that gray area, what's the requisite level? Look at the ingredients and just say, you know, how many dimensions is this really playing across how heterogeneous is this list.
It's not necessarily the number of ingredients, but what, how heterogeneous is is it is it and that gets into the entropy that ties back to category membership that ties back to, you know, uh Is it a nice clean mapping between solution and problem or has it got all these other things that it's trying to solve? And that could have some, some uh you know, problems down the road for, for whether it's health effects or the likelihood that it's gonna break over time.
So the more compressible, the more cohesive the object, the better. OK. Um So our quality heuristic really needs to satisfy, you know, generally defend the notion that less ingredients often do mean higher quality, but also handle situations where more ingredients could in fact be a better option and hint at the likelihood of an object bouncing out of its category, right?
Because at assessment time, at decision making time, at purchase time, that's all we got, you know, we don't get, you know, these are closed systems, we don't control the quality. Uh We have to have a way of assessing, you know, we're not gonna know if it's going to bounce out of the category or when we just want to somehow probabilistically assess kind of the likelihood of it bouncing out. And that relates to the, the entropy of the internal guts of a thing.
And that really comes down to the heterogeneity. And, and again, that, that quick kind of trick you can do that heuristic you can do is just think in terms of, hey, how many dimensions are actually at play here? I look at this cookie. It's got almond, flour, maple syrup, vanilla salt. But really that's two dimensions. That's material and flavor. I look at this other cookie. It's got flour, sugar butter, that's material, flavor and texture. So you know what?
The one with the lower fewer dimensions is less, less heterogeneous. And for, for all the reasons I discussed that's probably the better option. That's probably higher quality. Ok. Coffee tables, books, software people, right. Looking at a book and that book is doing a bunch of different things, right? You're looking at the chapters and maybe the chapters seem pretty cohesive. They seem to kind of subsume into the main title of the book.
That's probably good if the chapters are kind of all over the place or it seems really heterogeneous, I can't really compress it down to a few things, the, the cat, you know, just like the dogs and the cats example, right? If I can't get all those different chapters of the book to, to really cohesively, you know, in a more obvious fashion, relate to the title of the book.
Maybe that's not as good a book, maybe it's not gonna be a higher quality if someone walks into my office or I meet an individual and, you know, the things that they're talking about seem to align with what they purport to be, right? The title they give themselves or how they introduce themselves. There's a cohesiveness to that description. That makes sense if it seems really heterogeneous. Right. Again, not, I'm not saying a bunch of different things, not the size.
Although that could be the first indication, you know, someone's to do list is really long and they're doing a bunch of different things. But maybe if that to do list is pretty cohesive, then that individual is really aligned to, to, to very few things, they're only operating on a few dimensions, they're more cohesively aligned to the category to which they place themselves. I really just think this applies to everything, right?
Uh you know, software engineers will be looking at a code base, you might have a lot of code code, you might be doing a lot of different things. But is it cohesive, right? Which, which lends itself to maintainability and security and all that kind of stuff? Right? So coffee tables book software people, the things that we eat, the things that we consume when we have to make that assessment and we're not the ones in control of that quality. Think about how well you can compress, right?
The ingredient list of something, how many dimensions is it really operating across? What is the heterogeneity? And that's something physical we can latch on to. That's it for this episode. If you'd like to take a deeper dive on this topic, I write more technical versions of these discussions on both medium and substack you can find them at medium dot com slash nontrivial and Sean mcclure dot substack dot com. So, go ahead and check those out as always. Thanks for listening until next time.