Take your coin toss for example, but add a few more trials.
HTHTTHHHT is just as likely as HHHHHHHHH, but it is not significant in any way because if we toss a fair coin fairly, we expect a somewhat random distribution of heads and tails. That's what fair coins usually do. HHHHHHHHH, however, is not what fair coins usually do. It's what loaded or two-headed coins do. That many heads in a row has a much less than 1% chance of occurring, so we attach significance to it. You can show it easily enough with Bayes Theorem: Pr(E/H) >> Pr(E/~H) where E is the evidence of 9 heads in a row and H is the hypothesis "The coin is biased". As long as
Pr(H) isn't outrageously low (and two-headed coins aren't exactly rare), it will be more reasonable to believe the coin is loaded than fair. As the number of heads in a row increases, the confidence in the loaded coin hypothesis grows until it becomes irrational to believe a fair coin is being fairly tossed.
Please bear with me -- I've only had formal classes in first-level probability and statistics, though I can at least claim to have used those skills in other classes as well as in practice. My own interest in statistics isn't such that I pursue it much unless I need it for something, which has become increasingly frequent and will continue to as I advance in biology. Sadly, it won't be long before I actually have to
care what a Pearson-Spearman correlation is. I mean, I care for all the reasons I
should care -- I just don't like math.
OK, so I get what you're saying here so far as coin tossing is concerned, and I haven't (intentionally) disputed that. Of
course tossing 100 consecutive heads is statistically unlikely. Just in terms of p-value tests, if you set up a normal distribution in a histogram, where each bin is filled according to the ratio of heads to tails (a set of those permutations), you'll get a curve that makes this principle clear. The mode would contain the set of results where heads and tails are 1:1, the left tail end would have the results where the ratio is 0:100, the right tail 100:0, etc. Do enough iterations and the trend will be clear enough to show that 100 consecutive heads is most definitely way outside any acceptable number of standard deviations of the norm. While the probability of tossing a head or tail remains 50/50 for each individual toss, maintaining the streak becomes increasingly unlikely with a fair coin. Fine.
The problem remains, however, that your analogy still fails. Your point with this fine-tuning stuff -- and stop me if I'm misrepresenting you! -- is that the particular configuration of universal parameters is akin to this string of head results. But you need to make so many assumptions to justify that relationship as I explained in my previous post. In particular here, your assumptions are actually conflicting with each other!
The coin-toss analogy expects that a series of events happened. Universal parameters, however, we cannot describe as a series of events. They may all be interdependent. They might be fixed. We don't know. But the point it, the universe as we see it is better seen as a
single roll of a single die or the toss of a
single coin -- a coin that might be double-headed. In the absence of a series of events, the results of a p-test is dependent entirely on the probability of a given configuration coming up. With dice, we assume that each facet is equally likely and thus for any
single roll, any one result is equally likely as any other result.
I'll repeat again the major problem with ignoring the "if we're seeing it, it must support life" argument -- you're failing to acknowledge the possibility that the universe has already been rolled again and again and again and that the probably of this one occurring, the only one in which we can ever detect, has become likely.
Without more knowledge of the facets, we cannot make an honest statement about the probability of any given configuration. A better analogy is probably something more like what we see in chemistry, where the changes of seeing a given molecular conformation are determined by environmental conditions such as temperature and pressure, the presence of other compounds, bond strain, charge flow, etc. Under "standard" conditions, cyclohexane will be found most often in the chair conformation because it requires the least amount of energy to maintain, though other conformations are certainly possible. By all accounts, nature is driven by what is most
efficient, period. That's why rocks don't roll
uphill. I see no reason to assume that the foundation of nature itself would be any different, and I'll wager, neither do
you.
Because of this, I think you should consider a new perspective for your coin-toss analogy. Rather than seeing a life-supporting universe as 100 consecutive heads, consider that if there are normal constraints that make this universe more likely, than this universe might be more akin to a universe in which there is an equal distribution of heads and tails and that we're sitting in the mode.