So a couple of days ago, io9 posted a book excerpt on "how to think like Sherlock Holmes -- and have a better life". And it's great, as far as it goes. The bit they publish is all about "motivation and mindfulness", which is pretty much how you pick up the skill in the first place. You have to actually give a damn before you can learn anything. I've been known to tell people who ask how I do this stuff that I have "the magical power of paying some fucking attention," because that's the bedrock foundation of everything else that trick involves. (I also sometimes tell them "spooky psychic powers". This is because some people are superstitious idiots and won't accept another explanation. People who sincerely believe that magic is real and that science is wrong and evil because it doesn't know everything yet are surprisingly easy to deal with, if you can figure out how to tell them things in the proper mysterious fashion, and -- this is crucial, and sometimes annoyingly difficult for rational folk -- never back up your authoritative tone with any reasoning. Practice in the mirror, it helps.) Then you do it again and again forever until you start getting conclusions without having to consciously go through each and every step.

Lots of books make those points, and they're all equally right. The skill is something that can be taught to pretty much anyone who has a native talent for pattern-matching. It's... well, easy is probably the wrong word for it, but at least it's conceptually not complicated. They vary by method and in exactly which details they hit in their lesson plans, but the books and programs that give instruction in the combination of inductive and deductive reasoning Holmes refers to as ratiocination are mostly at least useful-ish.

They still piss me off, though, because they never cover the disadvantages. This one implies that thinking like Holmes will lead you directly to some sort of life improvement. I won't claim it's some kind of hideous burden -- I think it's mostly fun -- but there are a few pitfalls.

For one thing, there's no off switch. You get this skill by working at it constantly. The excerpt from io9 even says outright that the point is to practice until you can skip the part where you plod through the work step by step -- they note, correctly, that this results in what people call 'intuition'. It's not some kind of eldritch magic, it's just shifting most of the processing out of the conscious part of your brain back into sort of the liminal edges of your subconscious, thus freeing the explicitly thinky-brain parts for other work.

Ponder that for a second. You're trying to get the sherlocking to become automatic, which makes it, you know, automatic -- it will run all the time, without you having to trigger it on purpose. Which means always. Which means that once you pass a certain threshold combination of skill and stored context, you'll start to know things about people whether you want to or not. You will be unable to ignore things, even if that would be a lot more convenient. If your friends were dysfunctional when you started teaching yourself to sherlock stuff, they're still going to be dysfunctional when everything clicks and you get good at it, except now you'll know. And if the internet has taught us anything, it's that once you know something, you cannot un-know it, no matter how much time you spend trying to figure out how to bleach your brain.

For another thing,you're going to be wrong. A lot. They don't tell you that because it doesn't sound very impressive, but it's true. People will almost invariably give you the benefit of the doubt -- which sounds a lot like "confirmation bias" if you say it right -- as they do for Holmes in all of his various incarnations, and you can aid your good impression by glossing over the moments when your extrapolations went awry. But the accuracy curve for this is unfortunately asymptotic, and as you don't have a literal eternity or infinite brain space, no matter how good you get, you will never climb even remotely near a 100% hit rate.

How well you can get the sherlocking thing to work on people varies pretty widely. The pure forensic variant is actually the easiest one to learn, because when you're dealing with things like the laws of physical cause and effect, it's relatively easy to circumscribe the problem at hand, and thereby know whether you have all the relevant information or not. The more certain you can be that you know what variables will have affected what happened (or what's going to happen), and the more precisely you can fill in those values, the more likely it is that you can piece together a sort of world-line for the situation you're trying to decipher. Physical evidence has physical properties that can be measured, and follows the laws of physics, no exceptions.

People, not so much. I don't know from normal -- a perennial problem with me -- but I gather the variant of sherlocking talent that works on squishies is rarer than the one that works on trace. I base this on the fact that I don't generally discombobulate people like forensic anthropologists, but I have startled the hell out of quite a number of psychologists, both clinical and cognitive. I personally think that explaining it is a lot harder than doing it in the first place, but what do I know? If you have the knack, you still have to train it, which is where the lesson books are worthwhile. I couldn't do a damn thing with it until I quit listening to my mother, and started reading social psychology and picking up the explicit vocabulary for everything.

Being able to name all the variables that affect a situation requires you to know the entire context in which you're working, out of which said variables can be sifted. It is not possible for you as a limited, mortal human being to know all of the relevant context for another limited, mortal human being, much less all of the bajillions of people who have probably brushed up against your problem in one way or another. Not ever. The possible pieces of context jump from ω for knowing a physical situation down to an arbitrarily fine level of detail, to ωω and probably beyond for human beings. This means that whenever you go to finesse out the answer to a problem anywhere even remotely resembling the real world, you will not be able to work in classical certainties, but rather, in quantum probability clouds. If you've seen someone in a particular situation 20 times before and they've always dealt with it in the exact same way, they'll probably do the same thing the 21st time you watch them go through it -- but, then again, they might not. Just like the electron is probably zooming around pretty close to the nucleus of that atom you're squinting at, but might also turn out to be somewhere in Philadelphia. Any "answers" you get this way are necessarily arrived at on the balance of the evidence, not on cold, hard facts. Much as Holmes would like to pretend this is all black-and-white and completely logical, it isn't.

(Which is one of the reasons I like Gatiss' and Moffatt's Sherlock -- he admits to guessing. He's just guessing  really fucking well, taking into account a hell of a lot of factors that most people don't even notice.)

You will also, once you get good at sherlocking, be right an awful lot. This frequently garners an even more hostile reaction than when you're wrong. People are wrong about other people all the time. But there are certain areas in which most folk are wrong more often than they're right, and when you begin getting those bang-on, the people around you start feeling like you've gone rifling through the sock drawers in their immortal soul without so much as a by-your-leave. There's not usually anything in there that's terribly wrong or weird, but most people get pretty embarrassed if you start digging out whatever they've shoved into the back of their mental bureau, whether or not they're also afraid you'll start telling other people about it. My personal policy is to tell people nice things to their face and not mention anything unflattering unless they're asking for it -- either literally demanding to know what I think, usually under the narcissistic assumption that they are a special snowflake and I will be in awe of their true selves, or because they're being cruel to someone else, and a flat statement of the truth is the easiest way to interrupt them -- but if you have any qualms about people maybe thinking you're a goddamn weirdo, shutting up is probably a better policy.

And a final important thought, which might be especially vital information to the kind of science geeks that read io9: Knowing what is going on does not give you the ability to change it. This kind of thinking is particularly tempting for people who start out in the hard sciences, where generally once you know what the rules are, you can start working out how to bend them to your advantage. This is not the case with people. Frankly, knowing what is going on is mostly good for knowing when to fuck off with alacrity. (Don't knock it. This is a more useful skill than you probably realize, especially if you don't currently have it.) Once you figure out where all the option buttons on the world are, the temptation to press things in an attempt to make life better is almost overwhelming, but seriously, DON'T TRY. If someone asks for advice, feel free to give it, but always be prepared for them to completely ignore everything you say and head straight for the worst possible outcome like they're on rails. Arguing will just make them floor it. You will never, ever be able to talk sense into anyone who doesn't really want it. This includes, unfortunately, yourself.