#dohminion #dimonion #dominlon
He prefers to speak to Jeeps.
Take a look at this other part of the VA count as well:
I think you are both right. It appears fishy but is not impossible; it would be good to look at time points between the screenshots to answer the question.
If either of you really want to think about this type of thing in VA, this might be a more compelling case:
Are people archiving/cataloging these propaganda articles? They will disappear after we win as the cockroaches try to scurrying back into crevices.
Don’t get any sun, either!
cos tan = sin
(stolen from mathologer)
One of the things I keep pushing (in line with your concerns) is this super weird pattern in VA:
https://thedonald.win/p/11Q8Eb5Pcs/hey-va-lets-talk-fairfax-fuckery/c/
If this bit was “washed” it was washed with muddy water. It also does not expressly involve the imprecise vote share values (you can see screwy things from the full-precision total vote count changes alone). The vote share stuff is supplemental, though.
Lol, try yelling
I think you do it like this:
“wwwicegov” with no checkmark and 853 followers? Thanks, Jack; didn’t know you lurked.
I just donated my last $0.27; match me!
I don't know if it quite goes the way you are thinking. Here's the data with rounding:
https://i.maga.host/oZ1HIH6.png
Let me know what you think.
Although it looks like other people are already taking rounding/sigfigs into account in more recent posts, per your request I put together a script which accounts for sigfigs. It's pretty bare bones (a starting point), and I'm not a programmer, but if you want to tweak it or further develop it, here you go:
It will pull data for all states and output tab-separated lines for each NYT time point. Right now it just prints the change in votes with sigfigs for each candidate.
Borosilicate Johnson? What’s the story here?
Thanks. I’m talking about something different. We don’t have access to anything that shows what you are talking about (I don’t, anyway), but it’s good to be aware that it could be out there/in the machines.
Even if NYT was working with non-integer ballot counts, the lack of precision in the vote share values would still leave us uncertain of what the ACTUAL number of votes for a candidate was at any given time (even if the “actual” number was some screwy Dominion decimal).
Sure, no problem. By “cleaner representation,” do you mean with respect to the script/output, or something else?
I’m not blocking or even discouraging anything. I literally said that I encourage both the video and python script thing.
I’m pointing out relevant factual information: the script does not account for sigfigs. It’s still useful, but people might become unduly discouraged or lose confidence when they run into false positives. Or worse, they may incorrectly become overattached to a false positive from the script and start drawing incorrect conclusions.
Exactly, something screwy is going on with the x-axis there (15% movement from the same clump of votes being taken out and put back in despite being 9% in total). They also jump from 94 to 99% at 10pm on the 4th.
Yup, they are trying to be sufficiently non-transparent to prevent effective public scrutiny.
Oh, also: don’t forget about the possibility of false negatives from the Python script. (Less of a concern than false positives, but it may warrant a “second pass” through the videos looking at all timestamps in the NYT file.
Also, I’m on mobile right now, and I’m an amateur programmer at best, but it should be possible to write a script which identifies “confident positives” and “confident negatives”; it would just be a matter of taking sigfigs into account in the script (some of the “hits“ from the script are probably by such a large margin that the NYT rounding doesn’t matter, and these could be identified).
Thanks, that’s a good hypothesis. I like your thinking and that should be looked into.
I mention a couple other hypotheses here (scroll down to see the linked comment, not just the linked post):
https://thedonald.win/p/11PpTyOnqv/x/c/1ATVCpXsYt
Might want to check them out/spread them as well.
This looks great; very good to see. I want to make a quick PSA to clear something up about the Python script from the giant stickied post yesterday.
First, big shout out for having put that info together—I think it will definitely be useful, and this post about the videos is exactly why.
However, I think it should be made clear that the Python script generates some number of “false positives” (there are almost certainly plenty of true positives in there as well, so it is still a great resource to speed up this video analysis). The problem arises from rounded numbers in the source data from the NYT. I explain it more thoroughly below.
I say this so that you are informed and not disheartened if some of the “hits” from the Python script turn out to be false positives.
EXPLANATION:
I looked at the script. It uses the vote share ratio (at most 3 sig figs) and the vote grand total to calculate the total votes for Trump or Biden. So if the grand total at a particular point in time is 2,222,222 votes and Trump’s vote share ratio is 0.501, the script calculates it as 1,113,333.22. You can’t have 0.22 votes, so we already know the script is off. The problem is that the ACTUAL vote share ratio for Trump at that point in time could be anywhere between 0.5005 and 0.50149999... So the total number of Trump votes is somewhere between 1,112,222 and 1,114,444, but we don’t know any more precisely than that. That’s a range of 2,222. If you just assume that it must be 1,113,333, you could be off by as much as 1,111 for that time point. There will be occasions where the previous time point gave a high overestimate and the current time point gives a low underestimate, and when you subtract the overestimate from the underestimate you get a negative number (even though the true values might not produce a negative number).
Thanks. Had not yet seen that about fractional vote counting—fucking crazy indeed. Good to know. It is of course a separate matter from the fractional results arising from imprecise vote share ratios (I know you weren’t saying otherwise and were just cluing me in on the separate fractional voting thing).
I agree the there are probably true positives mixed in with the false positives produced by the script, and I do think even with all the false positives it is a good starting point to dig deeper. Looking at all hits identified by the script is much faster than looking at absolutely everything.
Would you mind taking a look at this post:
https://thedonald.win/p/11Q8Eb5Pcs/hey-va-lets-talk-fairfax-fuckery/c/
I think there might be something to this; could be a big lead/clue. Not dependent on the vote share ratio precision (3 sig figs are enough where discussed)
No problem! Keep getting the word out.
No problem, let me try to do a better job of explaining.
Basically the NYT data (used in that script) does not give sufficiently precise numbers (NYT rounds the Trump and Biden “vote share” value to only three decimal places).
Sig fig = significant figures. It’s a way of saying how precise a number is. The point is, because of this rounding, sometimes it will look like votes have been subtracted from a candidate (when that may not be the case).
Very interesting; thank you for bringing it to my attention!