Ilmari's Mini-Tournament #1: Rules

The rules are the same as for RMT round 4, with the following exceptions:

Mini-HSA is not included in the list of components; Only Carbonite and TimeScape are allowed.
You may not boot TimeScape, and it must be followed by at least 2 DATs. You may boot Carbonite, but the _executed_ copy must be followed by at least 8 "DAT 0, 0"s whether or not you boot it (ignoring here the possible presence of other warriors in the core).
Every player may submit up to 4 warriors. (even 4 copies of the same warrior if you really want) PINs are not allowed, and the definition of a p-switcher should rule out any other handshaking methods I can think of. Your score will be that of your best warrior.
The warriors will be pitted against each other multiple times, with N lowest scoring ones dropped off each time. The exact value of N will depend on the number of submissions I get.
Self fights will be included. The score in a self fight will be the average of those of the two copies (which means only the number of ties really matters).
The number of rounds per battle (pmars -r???) may vary. If you need this information, use the constant ROUNDS. (Feel free to use conditional code - it'll probably do you no good.)
Your program may not contain any STPs that are not actually used by your p-switcher. (I mean used, not just executed in some rare case for no real purpose.) This is to disallow brainwashing decoys.

Notes & clarifications

Both components must be included in your warrior, even if you only ever use one of them.
A p-switcher is allowed to delay its choice indefinitely or even to suicide. Such code is still not allowed to perform anything resembling an attack.
Neither is your switcher allowed to scan the core, obviously.
If you submit less than 4 programs, please mention explicitly if you want some of them to be entered several times.
You may submit more entries, or replace previously submitted ones, after your initial submission. Just specify clearly which entries you want me to use.

Benchmark warriors

I will submit the following 4 basic warriors as benchmarks; Naturally you may submit something similar if you wish.

AllC (Always run Carbonite)
AllD (Always run TimeScape)
DTfT (Carbonite if won, else TimeScape)
SoLT (Switch on loss or tie)

If you can't see any logical connection between the rules above, here's a hint: Think "Prisoners' Dilemma". It turns out that the scores of these two components approximate this well known game theoretical problem rather well.

Although no knowledge of the Prisoners' Dilemma, or game theory in general, is necessary for success in this tournament, it may prove valuable not only here but in normal hill play as well. While much of the game theoretical treatment of Core War strategies has centered on the intransitive nature of the game (the paper-scissors-stone analogy), the fact that CW is not a zero-sum game has major implications on the changes in hill balance.

While the Prisoners' Dilemma and its numerous variations have been extensively studied, there's still a lot to be learned in the area. In this case, the challenge comes from the random fluctuation of the scores and the limited information given, both of which make the traditional Tit for Tat based strategies difficult to implement effectively.

I hope that this tournament will help to provide better understanding of this aspect of Core War strategies and to provoke further study and discussion. And of course that you'll have plenty of fun playing!