On the relevance of data sampling rates to avoid aliasing...
The classic example is the wagon wheel on TV. If you watch the video of the wagon go by, the wheel sometimes appears to be turning backwards. That’s aliasing, and results from not sampling data at a fast enough rate to give a true picture of what is actually happening.
You sample some data (say a video frame of the wheel) and you see one state. The next sample (or next video frame) shows a different state but is it immediately after or is it sometime later? If the wagon wheel turns at five times a second and you take samples at two times a second you won’t get a good representation of what the wheel is doing.
Another example, say I am baking cookies and they keep getting burned. I’ve set the oven to 350 and I go to check the temperature with a thermometer every twenty minutes for two hours. I get temperatures of:
0 min. - 325
20 min. - 330
30 min. - 330
60 min. - 325
80 min. - 330
100 min. - 330
120 min. - 325
That doesn’t explain why the cookies get burned. If anything the data shows a constant oven temperature and the snickerdoodles should therefore be undercooked.
So I stop and think that I am sampling temperature every twenty minutes for two hours, but the cookies only bake for 10 minutes. I need to know what is happening in that ten minute window more than I need to know what the oven does over two hours.
So I check the temperature every 30 seconds for ten minutes and I see:
0 sec. - 325
30 sec. - 330
60 sec. - 335
90 sec. - 340
120 sec. - 345
150 sec. - 350
180 sec. - 355
210 sec. - 360
240 sec. - 365
270 sec. - 370
300 sec. - 375
330 sec. - 380
360 sec. - 385
390 sec. - 390
420 sec. - 395
450 sec. - 400
480 sec. - 400
510 sec. - 400
540 sec. - 380
570 sec. - 370
600 sec. – 370
So now I know why my snickerdoodles are getting burnt.
I sampled the same process each time. Both data sets tell me what is happening in the oven. But only the second data set has enough information to tell me what is actually happening in the time frame that concerns me.
If you look at discrete examples of behavior of a system over large time periods it doesn’t necessarily tell you much about the behavior of a system over short time periods.
Where vital records are concerned, I hope you can see that looking at three from August of 1961, one from May of 61, two from July of 64, one from January of 65 and four from November of 68 isn’t going to give you anywhere near the same level of useful information as would if a dozen records from a single week in October of 1961.
As for the Factcheck COLB, I think you are simply mistaken in your presumption that because you couldn’t figure it out it must be falsified. Especially since there are a number of photos from that stunt that do show the seal quite clearly.
You don’t seem to be listening to what I’m saying. What is observable on the Factcheck images is not reproducible in real life. There is no other angle that could produce the exact left-side edge shown on the Factcheck image. Every time you have the page so that the left-side edge looks like the Factcheck image, you get the same image of the folded “seal” - and it doesn’t look like what Factcheck has. The shape of the circle on the top fold is reproducible every time. The shape on the “authenticating seal” is never reproducible.
In science that’s definitive. If a result can’t be reproduced under identical variables, there’s a problem with that “result”.
Regarding the sampling of BC#’s, all I’ve ever used that for is to show that the numbers go up rather than down. That’s all the degree of precision that I was looking for. BC#’s from early in the year have low numbers, BC#s from the end of the year have high numbers. That’s all I was using those numbers for.
The numbers being serial is stated in the CDC’s Natality Report, and HI law requires that HI BC’s conform to the standards of the CDC. And the Nordyke BC’s are numbered serially, one right after the other.