Draft: Get Lucky

by atucker2 min read7th Aug 201236 comments

11

Personal Blog

Here's a draft of an article that I want to post soon, but I figured I might as well get some feedback before I go ahead. If anyone else has exercises or experience with this I'd love to hear about it.

The popular science article version of the research mentioned can be found here: http://www.richardwiseman.com/resources/The_Luck_Factor.pdf


There's no fundamental propensity to good outcomes, but it looks like Luck is a thing.

 

There were some experiments done by Richard Wiseman (of 59 seconds fame/recommendation) investigating luck, and they found that there was a statistically significant factor which led to some people being more likely to receive unexpected benefits.

 

He took two groups of people, one self-proclaimed “very lucky” and the other normal, and asked them to do simple tasks, like to count the number of times a photograph appeared in a newspaper. He didn't tell them was that he had rigged the magazine to have a few convenient conveniences, like a giant sentence telling them that there were 43 photographs, or large text about how if the reader pointed this out to the researchers they could get $250.

 

Still, to a statistically significant degree, the “lucky” people noticed them more. What gives?

 

Have you ever missed an important detail because you were focused on something else? Once you're looking for something, you throw out details that fit. If you're looking for things that fit into the steps of a plan, then you're more likely to throw out stuff that's not already included.

 

If you weren't trying to count the number of times a word appeared, you probably would have caught the sentences. Planning on counting makes you miss details like it already being done for you.

 

Engineers bump into this sort of problem a lot. How many times have you spent a long time trying to code something before you found out that it was included in a library? Or tried to write a piece of code that doesn't actually wind up being used because what it accomplishes isn't actually needed?

 

This failure mode is child of lost causes and priming. If you try to do X in order to get G, then you're primed to miss Y even if it's relevant to G. If the bottom line is written, then you're going to leave stuff out.

 

Insofar as Luck exists, it seems to be the ability to use unexpected but useful information.

 

Not quite sure how to practice this skill. A few things I've tried:

  • Meditation. There are a few mental operations that seem related to letting go of a set of predictions and looking at reality instead. Like, asking how it actually feels to be breathing rather than looking for what you think breathing is like.
  • Feel “open” to the possibility of something popping up. I used to find four leaf clovers this way – by walking around occasionally glancing at the grass, and not particularly looking for a four leaf clover, but trusting my brain to draw my attention to it if there is one there and visible.
  • You could probably actually walk around outside right now and try this. See what it feels like to notice details.

  • Walk around outside and see if any small animals catch your eye. Try to use the feeling of your attention being grabbed more often.

  • Look around the room you're in and see if there's anything that you've forgotten about, but is useful.

  • When given a plan, ask yourself what kinds of functions or adjectives something could have that would make it useful, without drawing a picture of what that thing is.

    • One time at a Rationality Minicamp someone asked if there was a sceptre-like object. Most people said no, and I said yes, even though I didn't know where it was. I was able to quickly find one in the room because I ran the algorithm of “find the cylinder-ish thing in this room” rather than the algorithm of asking myself what cylinder-like things there are, then go get it.

       

 

Personal Blog

11

36 comments, sorted by Highlighting new comments since Today at 9:02 PM
New Comment

Has anyone repeated wiseman's experiments? I'm suspicious of any weird one off result, especially in psychology where numbers are low and volunteers are required

Clearly we need a better word than luck

Perhaps...

"A stroke of luck" = "A surprising but useful contextual affordance"

"Lucky" = "Habitually contextually aware"

Instead of saying "Good luck!" perhaps we could say "Watch for opportunities!"

Opportunability

I feel like something like that ought to work, but that word is ugly.

Describe "lucky" people as "opportunistic" maybe? Opportunity-prone?

Is it clear?

People use "luck" to describe the ability to regularly have surprising and good things happen to them. If that exists, why not just use the word for it?

"Luck" is strongly associated with certain false ideas, even if it can be used in a clearer sense that doesn't involve them. You can't easily get rid of undesirable connotations of a word, so it's probably easier to pick a different word.

Any suggestions? If there aren't any good similar concepts, then staying with Luck makes sense.

"Truth" is associated with lots of false ideas, but we don't really have better options unless we want to go into Greek or something.

"Truth" is associated with lots of false ideas

The extent to which it may be so is not similar to that for "luck".

(If you can't find a better word, using a descriptive phrase without focusing on any single word is another obvious option.)

Using short words rather than short phrases makes a concept more salient.

All else is not equal.

[-][anonymous]8y 5

It can also refer to something better than you expected happening to you due to a statistical fluctuation (e.g. winning the lottery).

Go for one month saying Yes to every request or suggestion you hear/see (with some limits). It's amazing how much opens up.

If you want to follow up on the research, a good way to start is to pull up Google Scholar hits for post-1995 hits authored by Richard Wiseman mentioning 'luck': http://scholar.google.com/scholar?q=author%3ARichard+author%3AWiseman+luck&btnG=&hl=en&as_sdt=0%2C9&as_ylo=1995

And then you can pick key studies, click on their respective 'Cited by X' links (since presumably anyone replicating a specific Wiseman result will also cite said Wiseman result), and work from there.

I felt like this post was begging for a clever 'test' embedded within it like those "msot poeple don't notice letter rveersals" emails. However, I can't think of a convincing way to hide something in an article.

When I ran more D&D games via play-by-post, I wrote long campaign outline posts to get players, and in order to encourage people to read them I'd embed sentences like "PM me with the word 'chartreuse' and start with 50 extra XP" in long paragraphs. Not hidden, but a reasonable filter for people who bothered to read the things. (A lot of players were repeat applicants and learned to look out for these sentences; it didn't tend to catch newbies.)

I still do this when I write long requirements specifications at work, especially when they are supposedly being reviewed by people who I suspect will mostly just sign off on them unread.

So what you're saying is that your co-workers have characters in your D&D games. :)

I suppose.

Also, by the same token, that I write the most insanely boring D&D campaigns ever conceived by a human mind.

"OK, I implement requirement R1347-b in an Asian mobile deployment" (rolls dice)
"Oo, too bad. You encounter a previously undiscovered regulatory constraint that renders R1347-b moot. You fall through to R2218 and pick up an error condition."

Sounds more like a Paranoia game...

There were some experiments done by Richard Wiseman (of 59 seconds fame/recommendation) investigating luck

Adding a link to them here would make your article seem much more serious, and you could even see if they've been replicated and mention that.

Have you ever missed an important detail because you were focused on something else? Once you're looking for something, you throw out details that fit. If you're looking for things that fit into the steps of a plan, then you're more likely to throw out stuff that's not already included.

This paragraph looks wrong. I think you wanted "you throw out details that don't fit."

This failure mode is child of lost causes and priming.

I believe this should be either "is child to lost" or "is the child of lost".

Meditation.

I've typically seen meditation touted as a focus-increasing device, where focus is often the opponent of serendipity. Perhaps discuss that in the meditation recommendation? (That is, mention particular styles of meditation that will be helpful, as clearly some are. You mention enough details for someone who is already familiar to tell which kind you're talking about, but names or links to tutorials will be much more useful for newcomers to meditation.)

What I'm wondering is how people notice they're lucky. They'd have to know how lucky everyone else is in order to have a baseline.

[-][anonymous]8y 3

Huh, yeah, I had always thought that if Alice finds bills on the ground more often than Bob (all other things being equal), the most plausible explanation is that Alice is more likely than Bob to notice a bill if there's one near her, rather than that there happen to be bills near Alice more often than near Bob.

Given that, Alice will describe herself as "lucky", because she finds more bills on the ground. That same traits also makes Alice more likely to notice the text about getting $250 if she points it out to the researchers.

What I really want to know is how fast "lucky" vs. "unlucky" people would have counted the photographs if there was irrelevant intentionally distracting text on the page, rather than relevant text. My hypothesis is that "lucky" people would take longer than "unlucky" people, because they were attending to something irrelevant to the task.

A programmer who spends a lot of time looking at what libraries can do when he needs to be writing something that isn't included in them is unlucky- while a programmer who has already spent a lot of time looking at what libraries can do will be lucky when he needs an obscure function from one of them. But the chef who is distracted by C+ libraries when he burns his hand is also unlucky, as is the programmer who forgets a close-parens because he is thinking about what he will make for dinner.

Luck, then, is the skill of attending to each task the appropriate amount, when the appropriate amount cannot be determined from priors.

What I really want to know is how fast "lucky" vs. "unlucky" people would have counted the photographs if there was irrelevant intentionally distracting text on the page, rather than relevant text. My hypothesis is that "lucky" people would take longer than "unlucky" people, because they were attending to something irrelevant to the task.

Note that this is not a logical requirement: brains have LOTS of parallel processing, so having more stimuli be considered salient does not necessarily equal more time spent attending to the page. The amount of extra time (if any) would depend on the degree of stimulus discrimination that can be applied outside of conscious awareness.

That is, the lower the false positive rate that subjects' unconscious minds have in identifying opportunities for conscious attention and exploitation, the greater the tradeoff-free ROI.

Hence the 'further study' aspect. All I did was make a clear hypothesis that, if true, would indicate a further study or perhaps a direction in which we could direct 'unlucky' people in which they would be more successful than 'lucky' people.

I've started wishing people "Good skill!"

Fraudulent spam could be bad for luck. Claims of unexpected good opportunities that you need to learn to ignore.

I feel like I have read this exact article before.

PJEby claims that Wiseman claims not just correlation, but causation, that "believing you're lucky will actually make it so." Does he make such a claim? Does he provide any evidence for it? Under such a claim, it seems useful to call it "luck." But just under the claims you make, it seems bad to call it "luck," as many others have explained.

PJEby claims that Wiseman claims not just correlation, but causation, that "believing you're lucky will actually make it so." Does he make such a claim? Does he provide any evidence for it?

IIRC, Wiseman's book on his research said he investigated whether people could be trained to be "lucky", using the measurements he developed for luck (i.e., the noticing of unexpected opportunities), and that people cultivating a belief in their own luckiness increased their measured luck on his own tests as well as collecting their own evidence.

TBH, it's been a long time since I read the book, so I wouldn't swear to any of the above.

But just under the claims you make, it seems bad to call it "luck,"

Again IIRC, Wiseman defines luck empirically as something like "noticing and taking advantage of unexpected opportunities", which is reasonably close to the colloquial meaning, i.e. "having unexpected opportunities".

From a practical perspective, a person with Wiseman's luck (luck-W?) will perceive him or herself as having colloquial luck (luck-C?), and vice versa. Likewise the absence. So there is sufficient causal entanglement to not quibble over the word choice. For most people, It's easy enough to either:

  1. Convince them to change their definition, or
  2. Let them continue with their old definition, and increase their luck anyway by using Wiseman's methods.

There's plenty of precedent for both; we don't generally go around fretting over people using the word "day" without qualifying whether they mean a solar or sidereal day, or to double-check whether they're thinking the word has something to do with the sun going 'round the earth. For practical purposes, it's more than close enough.

Totally agreed with above. People talk about what they perceive, without noticing that different people perceive different things.

People know that some people benefit from unexpected opportunities, and call those people lucky. Lucky people notice unexpected opportunities, but don't know that others don't see them.

An actual statistical aberration after all other factors are controlled for should be expected to regress towards the mean, otherwise you haven't actually controlled for everything.

This is something different.

[-][anonymous]8y 0