# Blogathon: 12th Hour

Well, here we are at the halfway mark, and I’m afraid I’ve begun to develop a headache. A quite nasty one, too.

Also, it looks like we haven’t pushed much further towards our goal today in donations. Come on, guys. We should AT LEAST be able to break the \$90,000 mark this evening! If you’ve been following along with my blogathon but haven’t tried donated at least a wee bit to the SSA, I’m sad to inform you that this “no skateboarding” sign is very, very disappointed in you:

So I was mentioning how the darling, beautiful Benedict Cumberbatch was in my dreams last night.

As most of the geeky sorts who read my blog likely already know, Cumberbatch plays Sherlock Holmes in the BBC’s “Sherlock”, an adaptation of the Arthur Conan Doyle’s Sherlock Holmes stories set in 21st century London. It’s really quite, quite good.

One thing that always did sort of bug me about Sherlock Holmes, though, is the misapplication of the term “deduction”.

Deduction is a form of logic in which you determine a solution that is the only logical extension of the information you’ve been provided. For instance, if “anything that’s beige and rides a skooter is a Schmorpy” is a given, and you were then provided the information that “Dally is beige and rides a skooter”, “Dally is a Schmorpy” would then be a deduction, since it’s the logical conclusion based on the information. It’s the only possible, logical answer.

What Sherlock Holmes does is decidedly NOT deduction. What he practices is abduction. Or what is more commonly known as “guessing”.

He deals with probabilities. He observes his surroundings, notices salient or important information, and then GUESSES at the most likely explanation for why things are the way they are. But since they are almost never the ONLY possibly explanation, they aren’t “deductions”.

Now, this might seem like a silly little nitpick about semantics, but it’s actually something I consider extremely important to how we approach, and question, the world around us. Something that is actually very important to the specific kinds of issues I deal with.

You see, Sherlock Holmes could never actually exist and function in the real world the way he does in fiction because the real world is just too infinitely complex for those guesses, abductions, to be consistently correct or even roughly reliable. He would make mistakes CONSTANTLY… not because he couldn’t be that smart, but because the world is never that reliable in being probable. Sherlock Holmes could be every bit as intelligent as he is portrayed in the series, and could even be MORE intelligent than that… he could basically be a ludicrously advanced computer designed to constantly, instantly calculate the most probable explanation for everything around him and he would STILL be wrong too often to be reliable.

Improbable things ARE improbable. But when speaking of, well… everything around you, you get into “really big numbers” territory. In a situation reflecting a million different questions (and really, even the room you happen to be sitting in right now presents an ENORMOUS number of question, in terms of accounting for every observation that could be made about it and everything that’s in it), the probability of one of those questions having a “one in a million” explanation becomes quite likely.

All things considered, it would be far, far, FAR more improbable for everything to consistently have the most probably explanation than for them to quite often have improbable explanations. When speaking of the world itself, infinitely complex as it is, improbability is actually more probable than probability.

This is why coincidences can often seem to occur more often than they “should”. Out of the thousands of things that happen to you in a given day, a lot of them ARE going to be improbable in their explanations or how they work out or whatever. If you isolate a specific thing, improbable is still improbable. But when you’re allowing, well, EVERYTHING to be up for grabs, yeah, there’s going to be a lot of weird, chance things going on.

The first episode of Sherlock itself hints at this, suggesting a little bit of how the actual world works. Upon meeting Dr. Watson for the first time, and momentarily borrowing his phone, he “deduces” an immense amount of detail about his life. He guesses that Watson is a military doctor recently returned from the middle-east with a psychosomatic limp who is in need of a new apartment due to financial struggles, and is becoming desperate, since he refuses to accept help from his recently divorced brother due to pride and an ongoing quarrel between them over the issue of his brother’s alcoholism.

Sherlock gets everything right, except one detail. It’s not a brother, it’s a sister. She’s a lesbian, and the name “Harry” is short for Harriet. Sherlock remarks “It’s always SOMETHING!”

This is a small look at what a detective like Sherlock would constantly encounter in the real world. But it wouldn’t always be something. It would always be lots of things.

The reason this is important is because as we go through life, making decisions, allocating our attentions, time and energy, coming up with theories, trying to help people, whatever… we often base our decisions on a model of the world based on what’s likely, what’s probable, what’s common. But such models don’t reflect the actual world.

The actual world is a world of variance and diversity and improbable occurrences. A world of lesbians named “Harry”.

I’m a confluence of improbabilities. Transsexuality is rare. Green eyes are rare. Hungarian ethnic backgrounds are rare. Heroin addiction is rare. Intersexual characteristics are rare. Etc. etc. etc. But that doesn’t make me The Most Specialist Person Ever. It just means I’m isolating a few variables that ended up with improbable results out of the THOUSANDS of variables that comprise a human being. Most of which, for me, are nice and “normal” like “everyone else”.

And the thing is, most people are like that. Most people have at least a few rare or improbable traits. What would be TRULY rare is a human being comprised utterly and completely of only the most probable, “normal” characteristics in every conceivable way.

This is why normativities are so illusory and fleeting and inadequate. They don’t benefit actual human beings. Who they benefit is just a ghost. A weird, Platonic mapping of everything we, individually, swerve away from simply by virtue of being.

The only place there’s a “normal” is in our heads. Or, I suppose more tellingly, “normal” is only ever really embodied in contrast to who we really are.

If our models are to be meaningful, they have to be models that understand the nuance, unpredictability, and rich diversity of our probably improbable realities.

It’s elementary, my dear reeders.

### Comments

1. badandfierce says

Ever read Ursula K. Le Guin’s Lathe of Heaven? Not one of her better known novels. Not one of my favorites of hers, either. But I do really enjoy the way the profound strangeness of the protagonist is illustrated. He’s completely normal. Every personality test and psychiatric evaluation (not exactly careful research on the psychology front, but she’s never been a very hard science sci-fi writer, and maybe it was a bit more accurate in 1971?) places him exactly in the middle of the road.

Normal is totally weird.

2. Dendritic Trees says

I wonder if this goes some of the way to explain Holmes’ enduring popularity. Even though what Holmes does wouldn’t work in the real world, we like to hold onto the idea of the world as a relatively simple place, which can be understood by a non-specialist.

For that matter, one of the big draws of religion for at least some people is that the world as described in the bible (or other religious texts) is so much more simple and controllable than the world as described by science. The former says ‘this is so’. The latter says ‘according to our current best knowledge this is so under these circumstances 70 – 80% of the time, but not those other circumstances, which are different, and this fails to account for this new variable we’ve just discovered, which could totally change everything…’

3. A. Person says

A nitpick, but models based on frequency and commonality do reflect the real world, but are not accurate in regards to its totality.

If our models are to be meaningful, they have to be models that understand the nuance, unpredictability, and rich diversity of our probably improbable realities.

Models can be meaningful without mapping every nuance. In fact, overly specific models generate poor predictions and tend to be unwieldy to use. The trick is in realizing when a model is a bad fit to the specific situation. The problem then is trying to convince people that their worldview model does not fit in all instances.

• says

They don’t have to map all the nuances, but they do need to be able to ACCOUNT for them, and do need to not be totally dependent on the non-existence of the nuances and unpredictable variables in order to hold together.

• A. Person says

Natalie, in formal modeling the error term(s) accounts for nuances and unpredicted variables, but ultimately it is up to the person testing the model to determine if the model applies.

Models can be useful even if they don’t account for other factors, but in those cases they only apply in certain situations. Again, it’s up to the person to determine when it applies.

• says

That’s true. But I believe you’re talking about a difference of vocabulary: to a mathematician or a scientist, a “model” is a construct of assumptions, rules and probabilities about the real world; it is intrinsically logical, but the user understands that the model may not always be sufficient to predict an outcome in the real world, even when error terms are included. A human judgment may be required because of the model’s limitations. But to many people, a “model” is simply that which one uses to predict and describe the world, and thus may have components based on logic and probability, and other components like a judgment about whether the logical prediction applies.

The problem with Sherlock Holmes’ style of “deduction”, as Natalie has described here, is that he relies entirely on a probabilistic model, but his guesses are not incorrect as often as probability itself would predict. And of course this can happen because his world is limited by the author’s imagination.

4. sjrosewater says

There, you guilted me into it. I donated \$6.05 :), which was exactly the amount left in my Paypal account.

5. says

Just wanted to say, I’ve been following you all day, and I donated!

6. Vicki says

John Edward is the real life Sherlock Holmes. You don’t even need to be specific or approach 100% accuracy to impress people.

I have deduced that one of the hundred people in this room has a deceased loved one whose name starts with ‘s’!

7. says

I just watched the Sherlock films for the first time yesterday and I had EXACTLY the same thought about his use of the word ‘deduction’. And also how improbably it is that he’s right so much of the time… I guess I find it difficult to suspend disbelief 🙂

8. Joshua Adams says

I’ve got an omnibus of the Holmes novels/stories, and read several of them. It’s always bothered me how contrived the scenarios are that allow him to make his “deductions.” He often does it just as a parlor trick at the beginning of a story, where’s it’s especially obvious/obnoxious. There is always some piece of information that pretty much narrows the answer down to one thing (like a distinctive kind of mud being on a person’s boots which is only found in one part of the city, and there being only one destination of interest in that part–well, isn’t that lucky for Holmes?). While I didn’t read all of the stories, it didn’t seem like there was ever that “something” to trip Holmes up in the books. Everything just clicked into place perfectly all the time, and everyone was astounded. Nice to hear that the BBC series has a bit more self-awareness about the ridiculousness of the whole thing.

Speaking of the “Harry” anecdote, that’s pretty much lifted straight out of one of the stories, IIRC, except of course in the original it really was a brother, and Holmes guessed everything exactly, despite relying on far fetched inferences like scratches on the brother’s watch = unsteady hands = alcoholic. It’s just weird, and even maybe a little sinister*, how neat and predictable the world was in Doyle’s imagination, when you contrast it with the new series like that…

* (Of course it’s a brother! No one has a lesbian sister nicknamed Harry. Everyone’s the same, so you can reliably infer everything from superficial markers!)

• says

Actually, in Sherlock, in a “Study In Pink”, he DOES “deduce” Harry’s alcoholism from scratches around where the charger is plugged into the phone, indicating unsteady hands.

As though alcoholism is the only source of unsteady hands?

Aren’t nicotine and caffeine addictions a LOT more common?