ok

Take a left.

And don't give it back

Start giving a valuable piece of advice and then just stop in the

the left?

wouldnt it have been funnier to have just gone

Well, in all seriousness, I make some of my tactical decisions using a coin. I go through the options to find the best few, but often it comes down to an infinite cycle of 'my best chance is this, so my enemy will anticipate that, so i'll do this instead, but he'll anticipate that, so...". In those cases I just flip a coin to decide. That way I *know* he can't anticipate what I've done!

figlesquidge, game theory would be better. But I generally use that, and yes, it would be totally messed up by a coin flip.

Now, however, you've told me I shouldn't when playing against you. So ha.

I throw darts at a printout of the map till I hit places I can actually go. It usually takes a while. I'm not good at darts. :P

I like to attack one target until he's almost completely dead, then, let someone take over, while I kill that someone.

I can remember a game where I wanted to make a random choice of 2:1 probability (not in diplomacy or near a computer. I ended up saying 2/3 is about 21/32, and flipping a coin 5 times.

New players in a toss up will go for the supply center. Experienced players in a toss up will go for tactical position. Your tactically superior players in a toss up are unpredictable. =(

This works both ways. If they have to protect or capture, and they're a new player, you can bet they're going for the SC. If they're mid level, you can bet they're not going for the SC. Try not to fall into the same predictable pitfall.

Also choose the wackiest move when it comes along. They might see it coming, but they might be thinking... nahh... he'll never do it. Yes it could fail, but the sense of satisfaction at having pulled it off when it doesn't fail trumps everything else. =D

And the "holy sh*t" factor is not to be underestimated, either.

TheGhostmaker, you should have just used a three-sided coin.

you could use the constanza method and start doing the opposite of what your instincts tell you to do.

How to be unpredictable? Take a leaf from my PC's book. It just crashed, and chose to delete:

Every browsers' history, bookmarks, passwords et al.
The visual settings.
Every pinned item on the startup list.
Every desktop icon.
Half the commands to run programs at start up.
Every start menu setting.

In fact, pretty much every setting on the whole PC has been reverted to the default.

The thing has the cheek to welcome me saying it hopes I enjoy Windows XP, and inviting me to use "Settings transfer wizard" Stupid Moronic Sadistic Software!!!!!

Sounds like its time for a Mac! :)

One mouse button too few.

Also, had I a four year old mac that cost originally £400 including a screen, I doubt I would still be running it.

Macs can right click too...

Whenever I have a 50/50 choice, do I go for a SC or do I go for the position I often just toss a coin. Because I don't trust myself to be random.

When I'm at home I also use a dice (or two) to weight my options. Say it's autumn and I've got an Austrian army in Gal and Russia only has A Ukr to defend war. I might give Gal-Ukr a weight of 3/6 Gal- war a weight of 2/6 and Gal-Rum (or something else) a weight of 1/6 and then roll my dice.

And as Chrisp said, do something crazy every now and again.

i find i often have to think about how clever/stupid my enemy thinks i am, then i can guess his moves and act accordingly.

If you really can't find a dice, and you don't program and can't use a spreadsheet, and you want exactly 1/3 versus 2/3, toss coins like this:
If the first coin is a head stop right there
otherwise keep throwing until you get the same face twice in a row.
If you stopped on a tail, that's the 1/3 action.
If you stopped on a head, that's the 2/3 action.

Caviare??? What are you trying to say? Each throw has a 50/50 shit, so with you example, you get a 1 in 2 chance of getting the same face twice in a row (1 in 4 heads, 1 in 4 tails) and 1 in 2 of getting different faces (1 in 4 heads first, 1 in 4 tails first).

Crap, each has a 50/50 shot. What a Freudian typo...

Draugnar, the idea is that you throw coins repeatedly until you throw two coins with the same face in a row, EXCEPT if the first coin is a head. then you stop right away. So
H: stop right now (2/3 action)
T throw again
TH throw again
TT stop right now (1/3 action)
THT throw again
THH stop right now (2/3 action)
THTH throw again
THTT stop (1/3 action)
get the idea now?

Interesting method. It does add up. The heads would be 1/2 + 1/8 + 1/32 + 1/128 + 1/512 and so on which equals 2/3 while tails would be 1/4 + 1/16 + 1/64 + 1/256 + 1/1024 and so on which equals 1/3.

Start a thread called "what should i do to not be predictable" then ignore all the advice

What I've described is a simplification of a more general method for acting on any probability. Here goes, hold on to your hats:
First express the probability as binary fraction, lets take 1/7
which in binary is
.001001001001001001 etc
Now think of a tail as a 0, and a head as a 1.
Throw until your pattern of coin tossing differs from the binary fraction.
If you finished with a tail, take the action with the probability you expressed in binary. If you finished with a head, take the other action.
So all you need to be able to do is figure out these binary fractions in your head and your done. Simple :-)

This isn't a math thread, you NERDS. He wants to know what to do to be unpredictable.

I'll tell you what I do... I study the map a long time until I figure out the best move that I can make. Then, I figure out what is the worst possible thing that could happen (as far as the other players' moves) if I did the best thing. THEN, I figure out what would be the worst possible set of moves I could do to THEM, if they indeed do the worst possible thing to me. I do THOSE moves.

:)

Caviare, that is fair enough, except it isn't a proper algorithm, because there is no finish to them. My estimate of 21/32=2/3 is essentially the same thing, except I have said "And if it reaches the 9th throw, take option x

Ukla, the problem is that if you always go one stage down the reasoning, a good player will figure you out and go onto a third stage of reasoning. Also, you can find yourself looping around.

A1's best response is D1
A2 best defeats D1
A2's best response is D2
A1 best defeats D2

Just submit all holds. So many people use that tactic on this site. Sometimes they even do it twice in a row. Defenitly throws me off.

I know that the move circle is a descent into madness, so you have to stop yourself at some point. Understand, that I do what I said AFTER I figure out the best move for myself, which is the stage you are talking about.

Or, I'm just plain crazy. I'm not qualified to diagnose myself.

New Dip - that's exactly what I would expect from trim though.

Precisely, Ghost is right again. What you need to do in a number of situations is correctly guess how 'deep' your opponent is thinking, and by deep I mean number of iterations. Each set of best moves is based on the most likely responses, and these on the previous best set, a cycle that continues indefinitely

Ghost, the probability of the algorithm not finishing is 0. It will take an average of just 2 coins tosses to finish. Your point is philosophical.

Caviare, it isn't an algorithm, because it doesn't necessarily finish. Essentially what you are doing is the same as me, but instead of allowing up to 5 tosses of the coin, you allow infinitely many, which works, but for the fact that it can take forever (and at least a very long time).

I essentially used your method, but approximated:
2/3 = 0.10101 (base 2)

And that way, after five flips (1, 0, 1, 0, 1) You know it will end on a 1, so don't have to do the flipping (so you can stop for certain, and it is a genuine algorithm.

There are now 11 people whose name begins with diplomat followed by a number.

38 names start "diplo"

One thing I try to do to be unpredictable is to decide on some random criteria to do with word (e.g number of vowels) and compare the first messages from my likely allies to see which is closest to the criteria. I then concentrate on that alliance.

Ghost, the chance of the algorithm I described taking longer than say 100 tosses is 2^-100, a vanishingly small number. Its exponential decay! There are plenty of algorithms out there, probably running on your computer right now that aren't guaranteed to terminate in an absolute sense, but are in a stochastic sense and for all practical purposes. Because the pseudo random number generators used on computers have limited precision the algorithm I described, if implemented on a computer, is guaranteed to terminate even in the absolute sense you're talking about.
I already understood that your algorithm is an approximation to the algorithm I described.

True.

Also, for practical purposes again (I was doing this with a real coin), the accuracy of the coin is very significant. If it has a 51:49 weighting towards the 1, then that will soon be more significant that the accuracy of the approximation.

not follow any of the advice in this thread.