That's not quite right. Sid was responding to play testers reactions, not the vocality of community members.
He was continually talking about players, not testers.
1) Players and people in general do not have a good intuition about odds. But it's better for odds to behave how people expect, then how mathematics dictates.
I disagree, because:
a) This argument ignores the bad side-effects on people who
can deal with odds (they will choose suboptimal decisions because their expectations will be different from those of the target group towards which the game was geared).
b ) This argument does not take alternative solutions into account, for which these negative side-effects don't occur. Alternatives brought forward in this thread already are "Better information / helping the player to understand odds" and "designing the combat system in a way that the intuition matches better with the mathematics"
c) This argument reinforces a wrong understanding of odds, in one of the few games that actually has a reputation of teaching something useful once in a while. (As my spouse just put it after reading this discussion: "If I cross the street quickly now, I have an 80% chance of not being run over by that bus. Yay, I'm totally safe! *run*"). Okay, that's a slightly ideological argument, so feel free to ignore it - it might explain why for some people it feels extremely awkward to rig a mathematically sound system in order to meet false intuitions though.
2) The reason that surprising the player in this way is bad is because it makes the player step back and evaluate what happened. It's possible that a surprise feature has a net gain, but there's no real gain in this case.
The net gain is that the player re-evaluates his intuition and concludes that an 80% chance of surviving isn't as safe as he thought it to be. (That's actually exactly what I did after a few games of Civ4. I did have an understandings of odds before, but my intuition still made me attack with 3:1 odds when I shouldn't have. So I re-evaluated my intuition and as a result became not only better at playing Civ, it's also actually helped me in other situations too.)
If you don't see "understanding and learning how the game works" as a worthwhile net gain, then you could basically throw the progression of difficulty levels (that Sid also talks about in his speech) out of the window, because it uses the same process: Your powerful axeman slices through the AI's units. You see a chariot attacking and killing your axeman despite having a lower base strength. You are surprised. You step back and re-evaluate the situation. You learn that the chariot has an innate bonus which makes it a dangerous counter against axemen. You improve your strategy and start to look out for chariots when you march axemen against an enemy.
I'll skip points 3 and 4 because I think the whole spearman-vs-tank paradigms hinders discussion rather than helping it. Sid didn't mention it in his speech either. And the problem of high odds vs low odds is by no means limited to that paradigm either. I see little use in combining a general discussion about odds and intuition with an already incredibly loaded year-old discussion that represents only one special case of the topic in question.
5) The other reason that loosing with high strength differences sucks, is because it messes up plans. Being able to loose with high strength differences means that the outcome of combat is more variable; the distribution of possible damage is wider. This makes plans more likely to fail. It also makes plans more likely to go better then average, but the possible satisfaction of plans going better than normal does not possible disappointment of plans failing. This point applies to plans of both the stronger and the weaker side.
I think that this is a rather one-sided perspective. You can apply this argument in exactly the same way to any random element of any game that involves planning, so in effect you're advocating the removal of random elements in favor of more reliable planning.
This argument ignores that many players enjoy random elements, ask for them when they are not present, and/or even regard the ability to make plans which can handle unforeseen or improbable events as superior to make plans in a purely deterministic environment.
6) It's not a big change. The combat algorithm is mainly just less variable. And the computer is more readily willing to round near-zero numbers to zero. And combat is more in line with expectations in general. But that's it.
I agree that, seen for itself, it's a small change. I disagree that this is the only perspective under which it should be seen though. I think you're ignoring several implications of that change 8and the other advocated by Meier) which actually
are big changes. Three examples:
a) It changes the way the game is played. If 4:1 odds and above constitute a safe victory, then this does influence my decision of whether or not to attack a lot. This may be a desired change though, and even if it isn't, it can be balanced out by other changes in the combat system (letting defeated units survive may have such a balancing function). The change may not be bad, but it
is a noticeable consequence that arises directly out of the change that you label as "not big".
b ) So far, in the Civ series the game could always be trusted with regard to the combat calculations. In early versions of Civ, the match was simple enough to do it in your head. In Civ4, your chance of victory was displayed accurately (this accuracy of the calculation for Drill effects can be argued, but the problem there is that the effect doesn't lead itself well to a representation as a single percentage number, not that the game didn't
want to give you an accurate number). With the system that Meier advocated, the numbers can't be trusted anymore. In effect, Meier is breaking the "Unholy Alliance" (the term that he explains later in his speech) between designer and player by rigging a display of values that the players so far could rely on as being truthful. This is a rather big change in concept, and it casts a shadow on the reliability of other displayed values in the game.
c) If this change is the result of a new conviction that players are unable of grasping mathematical concepts, and incapable of learning (and that's unfortunately exactly how Meier presents it), then there is a definite danger that this conviction will have other undesirable results as well.
In short, I think you've fallen for the same fallacy as Sid. You correctly determine the way one group of players perceives and plays the game. But instead of thinking about the
other groups of players too, and trying to find solutions which work well for more than one group, you're limiting your discussion to only this single group. This is certainly easier to handle (in game design as well as in a discussion), but the groups of players who don't fit to the personality you're gearing your arguments and your game design to, won't be happy with that. That's why I said that (in game design) focusing solely on players who are unable to grasp mathematical concepts at the cost of players who can, will lead to unsatisfied customers looking for other products on the long run.