Assuming the whole for the part

We humans are interesting beings, since we’re both irrational and rational in equal parts (even if most of us deny that!). Again, this is an evolutionary trait, and we cannot dismiss the utility. We have already seen how important it is to work under the assumption that we don’t have all the data, and still manage to get the correct result, which used to be a survival trait. Thus, working with partial data is quite common for us; we talk about people who are “highly intuitive” because they don’t seem to need to reason much about something to come to a result. And often this result is purely irrational — in the sense that it doesn’t follow formal logic to reach conclusions — but nevertheless correct.

Let’s take our stupid examples again. Suppose you have talked to that group of movie fans an now have a good idea on the identity of the guy you’re dating, based on the reputation that this group has established for him. So instead of going to the French intellectual movie, you go to watch with him “Avatar”. After the movie, you comment that “Avatar” is already “the highest-grossing film of all time worldwide, surpassing Titanic, which had held the record for the previous 12 years” (see Wikipedia). Your date just comments: “I never watched Titanic“.

Now you’re baffled. Here’s someone who is a self-proclaimed blockbuster fan. His group of movie fans claim that he has seen every blockbuster since 1995. But he hasn’t seen Titanic. Something is surely wrong!

You might infer two things from this. One is that this guy is lying — either lying to you (he watched Titanic but for some strange reason he doesn’t want to assume that) or to the whole movie fan group (he is nowhere near the blockbuster expert he claimed to be and just managed to create a fake reputation by lying). The other is that the group you’ve consulted are absolute morons (they failed to establish a correct reputation for this guy) and not worth consulting any longer (in fact, their own reputation as a group has been severely diminished!). You might infer either or even both of them.

Statistically, however, it’s not very likely that this guy is lying — after all, as said, only 4% of us are (possibly) pathological liars. And worse than that, one thing is lying to a single person, the other is to be a convincing liar inside a community of experts on movies, all the time. Statistically speaking, this would be very, very rare. Not impossible, of course, but so rare as to be almost insignificant to consider. An alternative, of course, is that this group is full of liars that create “fake” reputations by covering up for their members (do your maths, this is even statistically less likely to happen).

So what happened here? Due to our excessive way of reaching conclusions based on partial data, we quickly fall into the Fallacy of Composition. This mostly means that we are prone to take the characteristics of a part for granted, and apply it to the whole. In the example above, we’ve fallen to this fallacy quite often. First, we assume that just because this guy never watched Titanic, he’s not a blockbuster expert as he claims (in reality, even a blockbuster fan might be unable to have watched all blockbuster movies since 1995). Then we assume that because the group of movie fans allowed this guy to lie once, they are all morons or liars (in reality, obviously, none of them has watched all movies exhibited since the dawn of time).

What happens is that reputation is also based on incomplete data, and, when we have only partial data to work from, this creates a huge amount of fuzziness. The group of movie fans did not see all movies, and they did not ask each and every member to prove that they have seen all movies. Your date has probably just mentioned quite a lot of them, and impressed the group with his knowledge. So the group inferred that he probably would also know some more movies that he watched but never mentioned in conversation, and that gained him credibility inside the group — since, of course, the group members are humans too, working with partial data, and also prone to the Fallacy of Composition.

You can see how messy this quickly becomes!

Here is my point, though. When validating a movie fan’s knowledge, which will establish their reputation as an “expert”, you might ask the following questions:

  • correctly identifies references in movies of a certain type that pop up in conversation
  • talks about movies that the rest of the group was not even aware of (i.e. showing that he has more knowledge that the average group member)
  • is consistently up to date with related/indirect information about movies beside the plot (e.g. knows a bit about the history behind the movie, or trivia about the actors in it, the Oscars or other awards the movies have earned, etc.)

But this might leave out the following questions that were never asked:

  • understanding of movies as an art form, e.g. the ability to discuss aesthetics
  • naming the soundtrack artists
  • knowing how much the movie profited in direct box office sales and indirect merchandising or DVD sales

So, once more, the tag “movie fan” is applied only to limited information that applies to a group. The last set of questions is not deemed “relevant” to be part of the “movie fans group”. Perhaps some other groups might find these relevant, but not this particular group.

What happens next is a question of expectations. If you are not a movie fan, you might think that all the above questions are equally relevant, and so, as a non-fan, you will judge a fan’s reputation by the ability to answer any of the above questions. That’s what you expect that a movie fan might know. In reality, however, it is the social group that defines the relevant norms. They validate their peers by agreeing on a common set of norms. Outsiders might have completely different perceptions of what the norms are, but in truth, these might not be part of the group’s norms. So your expectations towards recognising and accepting someone’s reputation is dependent on how well your own “list of possible norms” fit within the group’s own existing norms. If there’s a good match, you will be highly likely to pass along that individual’s reputation; in fact, if you are already a member of the group, you might have perfect knowledge on the group’s norms to evaluate reputation, and thus not be disappointed if an individual in that group doesn’t know how much the movie made in DVD sales. As an outsider, however, you will always project your expectations towards the group: you’ll get disappointed if an individual doesn’t conform to your expectations of what that individual is supposed to be!

Back to the real examples, this is where the Fallacy of Composition plays against us. In the real world, for instance, Governments will allow a company to incorporate if they follow a set of Government-mandated criteria, which will be validated. Thus, in the “social group” of all companies incorporated under a specific government, they will all conform to those criteria: they will “follow the law”.

The law will mandate a lot of things, like, say, silly examples that every company will need to have a fiscal address. Let’s take that as an example. So, every company incorporated under a specific government that forces companies to have fiscal addresses will know that every other company incorporated under the same government will have a fiscal address, too.

What about a company incorporated under a different government? Well, you will expect them to have similar laws, too, and so you project this expectation that they will also have a fiscal address as well. When suddenly you find out that they do not have a fiscal address (because their own government doesn’t have that as a legal requirement), what happens? You mistrust that company. They’re lying. They’re dishonest. They’re cheating. They must be breaking the law.

But that other company might send you a credential from their own government claiming that they, too, are fully legal and validly incorporated under that government. At this point, you cease to “believe” (or trust) on that other company’s proxy (i.e. their government) as establishing legitimately incorporated companies. You don’t wish to have anything to do with them.

The important thing at this point is to understand that no matter how many “proofs” this other company will send to your offices, no matter how many legal documents they send, you don’t trust their reputation, because they’re working under a set of social norms that are not familiar to you. Again, this is the fallacy of composition at work, linked to the notion that we make very quick judgements based on partial data.

Of course, I’m well aware that the requirement for a fiscal address is quite a silly example :) But let’s try with a much harder example. Let’s assume that for a specific government the set of laws for incorporation insist on a fiscal address; naming the partners; naming the jurisdiction in case of conflict; paying taxes annually. You can safely assume that the government will ensure that all these conditions for incorporation will be met. We can even imagine that this government will disclose the registries where they can prove that company X is, indeed, validated and certified by government as being a fully legitimately incorporated company, and that the criteria have been successfully met for several years.

Now you wish to buy some services from company X, say, in the multimedia industry. Will you trust them based on government credentials on their legitimate incorporation?

I’m prepared to claim that 96% of you will immediately say: “yes, of course” :) This is because almost all of us will immediately fall into the Fallacy of Composition again. We will intuitively think (but not logically!) that a company that is in good standing with the Government, in the sense that they have been successfully validated to meet all criteria for incorporation, is automatically a good multimedia service provider. “Good” in the sense of being able to provide service as claimed; to be able to fulfil contracts and obligations; to deal honestly with payments; to provide a service of quality.

But why we think like that? Governments never claimed anything of the sort.

By contrast, you might engage in conversation with the “social group” where this company works. That means “getting references”, i.e. talking to clients and employees of the company, see if they have a good track record in the industry, and so forth. This will establish that company’s reputation in the area they’re supposed to be providing services. They might be part of the local trade association or something like that, and you might be able to ask them what they think about this company’s service. In effect, the company’s reputation is established not by Government (in our example!), but inside the group of interactions with clients, partners, and suppliers.

Still, if you don’t know where this “group” is, what do you do? You rely on Government-as-a-proxy to establish reputation. We do this every day. When we buy something on the grocery store, we haven’t asked our neighbours or the grocery store’s suppliers and partners if their fruit is fresh. You rely on assuming that if they’re open for business, it’s because the local government has established some laws to regulate the market where this store is operating, and you can safely assume that “selling fresh fruit” is part of the regulations. You totally assume reputation-by-proxy.

Please note that I’m not criticising this way of thinking. We do it every day! In fact, I’m prepared to admit that we couldn’t run our complex society any other way. And, of course, I’m oversimplifying things — depending on the nation you live in, the actual set of regulations is far more extensive than I’m implying. Being part of a local trader’s association or a Better Business Bureau might be compulsory in some legislations. Some countries might force you to display on your grocery a list of certifications from quality institutes that regularly control the freshness of your fruit. In some cases, all public complains against the grocery store might have to be published publicly for everybody to consult. There is no universal rule — we don’t have a planet-wide government! — but it’s safe to assume that the “social rules” for grocery stores is quite more complex than I showed on these paragraphs.

Nevertheless, ultimately this might not answer the question: “are the shop attendants nice?” (which might be a criterium for you to go back again and shop there!) Even “is the fruit fresh?” might depend on your expectations; legislation might define it in a way you disagree with.

The beauty of the system is that 96% of the grocery stores you’ll find will very likely match your expectations :) (yes, again, only 4% will lie about what they sell). That’s good enough for most of us. We’re prepared to eat that odd rotten fruit bought at a new place when we’ve already bought 25 good, fresh ones every day there.

The dark side of the system is that just because in 96% of the cases we will blindingly accept reputation-by-proxy, we’ll still be very wrong on 4% of the cases. And, more dangerously, we’ll expect that good reputation-by-proxy on a very specific area will automatically mean that there exists good reputation on all areas. In my example above, just because a grocery store might have a fiscal address (good reputation which is validated by Government as to compliance with local laws), it does not mean they’ll have fresh fruit too. We just assume it does. In 4% of the cases, we’ll be wrong. But that’s not only because the grocer is a liar; it’s because we’re transferring reputation-by-proxy from an area to another one which works under different rules and assumptions.

Thus, on the reverse example, you might ask a group of fresh fruit fans where to shop, and they’ll tell you which shops to avoid, based on a criterium that’s good for you: fresh fruit. On the other hand, that rickety stall quickly assembled near to a corner of the street with the dubious character running it might have the most lovely fresh fruit you’ve ever seen in your life… but very likely have no fiscal address. Which is more important, compliance with the law (one social group which might have little relevance to your health or pleasure when eating fruit), or reputation from the group (a different, unrelated group which has little interest in how grocers pay their taxes, but loves healthy, fresh fruit)?

And that is a question I can’t answer easily :) So it’s time to keep this in mind, and switch over to the service economy in the 21st century, which is digital and online.