How I used to dread those fat envelopes....
And like my mother, you are making quite extraordinary
assumptions about me.
But let me take it in bite sized chunks, and preface them by saying that no, I am not forgetting the things you allege I am forgetting. You are drawing a vivid picture by connecting a certain set of dots. I, to be honest, think that there are some important dots you yourself are leaving out, because they would change your picture (or at least make it more nuanced) and I think that some of your dots are spurious (yeah, I can do metaphor too....)
OK:
First off, you mistake an assumption for a hypothesis (remember I am a scientist.) I do not
assume any of the following:
--that the fraud would be easiest to detect in DRE precincts
--that the fraud would be easiest to detect in any comparisons with Bush 2000 (different voting system entirely)
--that the fraud would be easiest to detect in swing states(*)
--that the fraud would be easiest to detect in Republican-controlled precincts
--that small inconsistencies are not meaningful (i.e., cross tabs)
What I proposed were testable hypotheses. To understand this, consider what your response were be if these were tested, and indicated that, for example, redshift was significantly (statistically) higher in DRE precincts than in precincts using optical scanners (or vice versa - there are good reasons for using two-tailed statistical tests). Failing to find such a correlation proves nothing. However, finding such a correlation would be extremely suggestive (although we have to be on guard for innocent collinear variables). This is why I am constantly on the lookout for testable hypotheses, because while, mostly, failure to that a particular effect is "significant" simply means that your precise hypothesis was not supported, finding a "significant" effect may well consititue important evidence.
So to consolidate so far: these are NOT assumptions; they are a list of hypotheses worth testing. At least one appears to be supported - redshift appears to have been greater in swing states. This is suggestive, although many interpretations are possible (including a rather boring mathematical one I won't go into here). And BECAUSE these are not assumptions, you need to back off a bit (as I would have said to my mother) because I am not up to what you seem to think I am up to, and in fact, am suggesting potentially fruitful lines of investigation - or at least soliciting more.
Second off:
You accuse me of not considering:
1. fraud "cleverly randomized to avoid statistical detection"
Oh yes I am. I don't expect you to read my posts particularly avidly, but if you had been you would have read the many occasions in which I have made the point that, perhaps contrary to lay intuition, random fraud would have been much easier to detect than non-random fraud. Random fraud is actually difficult to square with the data as we know it; however non-random fraud (fraud specifically targetted to hide the patterns that "randomness" would create" may well be consistent with the data.
2. "and that small percentages were stolen here, there and everywhere, with some concentration in the swing states"
Well, yes, and contrary to your implication, this is precisely what I have suggested, in several previous posts. Ask Land Shark.
3. "and that these thefts occurred in the CENTRAL TABULATORS, not in the voting machines (the DRE switches might even have been a red herring)"
Well, there is a bit of a problem here: I agree that the central tabulators may have been used for fraud, but that wouldn't account for much of the precinct level discrepancy observed in the exit polls, as the precinct level discrepancy was largely calcuated on precinct counts. But again, if you had followed my posts on this subject, you would know that this is a hypothesis I have been keen to explore.
4. "small randomized shaves (of various kinds: switch from Kerry to Bush, switch from Kerry to 3rd parties, 'disappeared' Kerry votes, added Bush votes, etc.) over a wide grid of results would be very hard to distinguish from the little up and down blips of any exit polling, and could only be seen clearly in the aggregate--a 3% margin of victory to Kerry in the exit polls, overall."
Actually about 6.5% but whatever. And yes, you would see it in the aggregate. I would also argue that it is difficult to envisage a scenario (and believe me I have tried to model several) where "these small randomized shaves" - if they were responsible for a substantial proportion of that 6.5% (insert: difference between margins) aggregate - would not also show up, aggregated, as a positive correlation between redshift and swing, which they don't. Either they weren't random, or they didn't add up to the redshift.
Third off:
You take me to task for my interpretation of the swing state finding, which a bit below the belt, as I actually pointed it out, and placed it in the "supports fraud" category, merely footnoting, as any good scientist should do, that a reasonably plausible alternative interpretation is also possible. This is QUITE DIFFERENT from arguing that the alternative is more likely, or that because there is an alternative, one shouldn't consider fraud. It is merely doing what good data analysts should do. I chalk the swing state finding in the column marked "consistent with a fraud explanation"). I would like to keep it there please, asterisked as necessary.
The rest of your post I won't address in detail, as it is a full-scale rant (and I love rants) not specific to me. I simply disagree with you about the pollsters, but I have addressed that point elsewhere. But I will end by saying:
I do not believe, in any circumstance, that "he who is not with us is against us". I may not be with TIA on his interpretation of the numbers. I may not be with you in your interpretation of more qualitative data. But this DOES NOT MEAN that I am against you, or what you want to achieve. I applauded an excellent point you made upthread. Frankly, ANY disenfranchisement is QUITE UNACCEPTABLE in a democracy, particular one that wields the power that America does over the rest of the world IN THE NAME OF DEMOCRACY.
But I happen to believe that bullet proof arguments are more convincing than flawed ones, especially when it comes to quantitative data. And, more importantly, I think that a well informed look at the quantitative data, can lead to real progress in finding out what went wrong, whereas I think that blanket assertions, while they may be great for inspiring activitist to action (and as I said, I like a good rant) can actually lead people in the wrong direction. I think (after extensive consideration) that the exit polls are the wrong direction. You are entitled to disagree.
But, like my mother, you seem to have got completely the wrong end of the stick about where I am coming from, or where I want this campaign to get to. That may well be my fault. But I have at least attempted to put the record straight here.
Edited to correct: Sorry PP, you are quite right with your 3 points. I am so used to thinking in terms of difference between margins at precinct level, I thought you had underestimated the discrepancy....)