I read this a few days ago and I'm sure it will appeal some L19ers. After reading it, I picked The Dice Man as light read. Was weird.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Interesting read, but it had the complete opposite effect on me.

[admin]
The rest of this post was randomly chosen for deletion to see if it improves the thread.
[/admin]

I've won two games where my opponent has resigned immediately after my misclick. So maybe there is something to it ..._{or more likely it says something about the quality of my game.}

I've never read The Dice Man, but some applications of the use of chance during World War II are discussed by Steven Brams in Encyclopaedia Britannica's article on Game Theory, as I recall from commissioning and editing the piece. A couple decades ago I got slightly interested in genetic algorithms. Have you read John Koza's Genetic Programming? There has certainly been more recent work on applying the theory of Markov Chains (random walks) to computer visualization, with some success in real-world applications, though I do not keep up with this field. Or if you have more interest in life sciences, I highly recommend Stuart Kauffman's The Origins of Order: Self-Organization and Selection in Evolution, which contains a great deal of information on the benefits of randomization on adaptive selection in complex systems. (Yeah, I'm running out of rooms to store my books in...)

As for go, it is apparent that rank beginners play better than random, but after they have learned enough, they do not always consider the best plays. Perhaps at that point they could benefit from a bit of randomness.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I'm going to try this out playing Go. Not just complete randomness, of course, but those positions where you've got a couple different viable looking paths available to you: Do I expand my moyo, or reduce his position, or maybe invade that corner? Or: what's the proper invasion point here?

It does seem like us weaker players sometimes get too attached to particular sequences. We know the 3-3 invasion of a 4-4 point, so we rarely try 2-4 or 3-4. I know something has worked in the past so I repeat it, unaware a different, better choice exists. Resorting to randomness when I like one move better than another but can't convince myself it is objectively better sounds like a good way to stay out of my comfort zone.

The real problem is not what to do if you can't choose wisely, but what to do if you seem to be able to choose wisely. Since we do not know anything for sure, it doesn't matter what we do about it, we can always fail and since we are not in control over that it's hard to prefer reason over unreason or vice versa. We're stuck in indecidability. E.g. if a pro opens with 2-2 and 2-2 then all experts might say that this is very bad and he might lose the game against his pro opponent. But then we all have seen those games where surprising moves confuse the opponent and sometimes turn out to be the exact moves that saves the game later. And if it works once, it can works twice....

So it seems: The deeper we reason about things the more we come to the conclusion that reason is a "fuzzy" thing, just like the deeper we zoom into the world the "fuzzier" particles get. Maybe we are outsmarting ourselves here and we should stop digging too deep.

I may add, that I wasn't totally thinking of this article as related to go (hence, Off-Topic.) I'm currently randomising my to-do list in several ways, as a "test" of this. 90% of what goes in my todo is not time-bound or urgent (or more precisely, all is urgent whereas some things are incredibly urgent,) so picking a next task can result in some task not being picked for a long while... How I'm approaching this randomised to-do? (also a glimpse into what my day looks like)

First, choose randomly among meditate, n-back training, anki (I've sadly gotten really back into my anki training program after a post-stress week where I did "nothing", so now I'm ~300 cards behind each day and don't really have the 45-90 minutes or so to get them to a manageable size, so, little chunks and patience) or "other." To do so I mentally order the tasks (like this would work) and then generate a random number between 0 and 1 with an iPad calculator. I just bucket tasks then (0-0.24 is task 4, 0.76-1 is task 1.) This is easier than using a d6 (6-sided dice), specially if there are 5-7 tasks (you can do some tricks if there are an even number of tasks, to get a uniform distribution among them with a dice, simplest case would be 12: throw dice, if 1-3, you pick between tasks 1-6 with the next roll, if 4-6, among tasks 7-12.)

During breakfast I enjoy some light reading, instead of wondering what book to read (my to-read pile is 20 deep or so) each morning and just settle for reading HackerNews on iPad, I set a list of 5 books and 1 category "other" (which I then reshuffle into Instapaper stacked reads, pending links in Mobile Safari, PDF books in GoodReader, in this order) and pick at random.

Then (these days, at least) I'm working on my PhD. I'm finishing the last section (some proofs still incomplete or unclear) and still revising all the rest. I choose among "keep working on pen-paper->TeX," "check section blabla for theorem blablabla" (a specific task I set) and "check complete random section" "check random lemma/theorem in random section". If I feel specially tired I add the "Russian roulette" option: Fun. If fate picks fun, I choose a task from the fun list. Then, I roll a random 0-1 for duration, with min 15 min, max 90 min. So far this has worked great: I've steadily worked in it all workweek days (except today, I was sick all morning and I barely worked for "work"...)

When I'm happy with the PhD work done (usually 1h30 of real work done) I turn to the work to-do. Usually I prepare a to-do of all that needs to be done "soonish," including work that needs to be done on the day, add "Fun" and copy-paste the list (I use Clear for my todo) in emacs, where I randomise the list (select list, M-x randomize-region.) Depending on how I feel, I either choose a random duration, do the full task or just accomplish a "chunk" of it.

Another part where I'm enjoying the randomness is idle time where I don't feel like pursuing complicated stuff and I just slouch in the sofa while playing something relatively mindless (no go, no reading, no-nothing) iPad game. Usually I just pick the same game I picked the day before, so, mostly I keep playing the same game for a long while, specially if it is very mindless and I'm very mentally tired. To put it in context (before I started this process) I had exactly 90 games. Most of them I had only playtested for review purposes, some others are what I consider "classics" and I enjoy having for occasional bursts of fun. Now I have been randomising the choices, and I have enjoyed some games I never played properly. This has lead to discovering I enjoy backgammon a lot (fate dictated I play a backgammon game I had on Thursday, liked it so much that I got a proper, decent app for backgammon on Friday and since then I've played almost 200 games already!)

So far, I'm liking this randomness. It's keeping me out of my usual ruts, and also eases my doubting moments "what to read" and "what to play."

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

On times when a random guess would have been a disappointingly good idea:

http://snarxiv.org/vs-arxiv/


Tangentially, moving every year is a quote-unquote good idea in slash and burn farming, right?

_________________
Revisiting Go - Study Journal
My Programming Blog - About the evolution of my go bot.