Or, for that matter, why didn't Sklansky, who was at the time no doubt talking a lot with him in their collaborations?

Actually, 'invent' is not the right word -- what I mean is, why wasn't Malmuth the first person to see that ICM could be used so effectively to analyze tournament play?

He presented the ideas behind ICM in his 'Gambling theory and other topics' which I believe was published in the 80s but he did so only as a way to determine equitable deal making. (the sections containing this info I believe is just titled 'Settling up in Tournaments' and talks only about how to calculate a fair chop.

The term ICM was coined by Bozeman in 2004 on the one table tournament forum here, and in reference to the GtaOT work. Bozeman stated at the time:

And it was Bozeman, and some of the One table tournament forum at the time who really pushed the idea that not only could this be used to accurately determine $EV from chip EV, but that this conversion would be a great tool for hand analysis --The exact same kind of hand analysis being done 20 years earlier by malmuth and Sklansky with just straight $EV.

So, if Malmuth had the tools to convert Chip EV to $EV, and he and Sklansky were regularly doing $EV calculations for cash games, how on earth did they not see the potential for analyzing tournament play?

Holding on to that one for a while huh?

It is not the same making deals after last hand has been played (by transforming chipstacks into % of prizepool for all players left in tournament) and analyzing possible decisions in hands beeing played (to push, resteal or call). It is probably most accurate in preflop allin decisions...and this is bread and butter of stt. I think they werent talking about decisions which could be make in particular hands.

In example...3 players left in the tournamet, sb pushes and player in bb has a decision to make if he should call/fold based on chipstacks of players left and prizepool. Dealmaking based on ICM is made before cards are dealt.

...or it is completley wrong??

Right, but still, they had all the tools.

They knew how to convert the stacks to $ equity estimates and they were regularly doing EV calculations (including all-in calcs). It's just amazing to me that they never put these ideas together (maybe they did).

Though you do make a very good point -- the preflop all-in is the bread and butter of STTs. Maybe they knew that it could be done, but never before had a game placed so much importance on this one aspect of poker. Still, tournament poker in general does require a lot of preflop all-ins once the blinds get relatively big. Sklansky, in fact, cooked up his famous all-in or fold system (published in TPFAP) so he had a pretty good idea how powerful the move is and must have known that it would be useful to have $equity estimates for use in hand analysis.

And if they knew (for example) that a particular preflop all-in (make up whatever works) with a $1000 stack was (say) +$20 cEV, it's just amazing that it never occured to them to convert the starting/resulting stacks with ICM to determine if it was, in fact, +$EV

I'd really love it if Mason or Sklansky would comment on this and whether those ideas were ever bounced around but were deemed too calculation intensive for the potential analytical gain looking at live MTTs at the time.

I'm moving this to poker theory.

At this point it has little to do with STT strategy.

I was recently asked about the origin of the Independent Chip Model and just discovered this thread. The answer is that I first demonstrated it in a 1986 article written for the old Poker Player Newspaper. Then it appeared in the first edition (1987) of my book Gambling Theory and Other Topics.

For those of you who have the latest edition, 2004, of my Gambling Theory book, you can turn to page 199 and start reading. There you'll see the first writeup of the Independent Chip Model as I use the method to determine whether it's correct to rebuy in a percentage payback tournament.

Best wishes,
Mason