10 Comments
User's avatar
NC ForSubstack's avatar

It sounds to me like your innate ability to see and recognize intricate patterns allows you to identify perhaps secondary and tertiary algorithmic patters while the aaaaaverage Bear (like Yogi! (and people like us)) have trouble spotting the primary pattern even if it is staring us in the face! Again, God bless you, sir -- for I find your talent amazing and a Godsend! Thank you again for your diligence and altruism. 🫡

Expand full comment
NC ForSubstack's avatar

You have been equipped to become the expert on this, and God bless you and even increase and enhance your knowledge.

Expand full comment
Jonathan Carrier's avatar

Wow! Very strong comments and replies. I can't add.

Expand full comment
Alex Tsakiris's avatar

So Andrew might this play out in terms of the legal action you've told us about? I mean, this seems like a slam dunk from a statistical/probability standpoint, but our legal system adversarial... how's the other side gonna push back? Do you anticipate being called into court as an expert witness? Have you asked AI to Steel Man this?

Expand full comment
Andrew Paquette, PhD's avatar

I have been asked about being an expert witness by a couple of different parties. Fees were discussed, but that's as far as it's gone so far. As for using AI to "steel man" this: this is the first I've heard that term, but if I understand it (same as "red team"?) then yes, I have. The AI thinks the WI evidence, as described in my paper and here, is so strong it is irrefutable proof of design for some covert purpose. According to the AI, these characteristics of the data cannot be artifacts of normal usage or innocent mathematical artifacts of any kind. It went from saying the probability of an innocent explanation being so close to zero that it was effectively zero, to "the probability is zero" after I pointed out the S1 algorithm is present in all 7 snapshots of the DB that I've seen, despite differing numbers of records in each, and a shifting starting point for the first S1 record.

That said, I know AI can be misled, so in my mind, there may be a counter-argument not contemplated by AI. Speaking of which, I did ask it to come up with and do a plausibility check of every explanation it could come up with. It came up with about 20, all of which it ranked at zero or near-zero probability.

Expand full comment
NC ForSubstack's avatar

Did you write a substack article about that particular AI use with the WI evidence? (I don’t recall putting the link to any such article in my collection of your other similar articles at https://tinyURL.com/ZarkAsksAI and I will add it there if you wrote one.)

Expand full comment
Andrew Paquette, PhD's avatar

I don't think so.

Expand full comment
GAVEMartin's avatar

Dang! I am not smart enough for this! I thought about it all night. Why is this so complicated?

In my mind's eye, I am seeing a player piano. The voter roll is the scroll on the feed. The perforated holes are a candidate's or a group of candidates' votes. Each race or group of races plays the same song...I dunno. Which would mean, even worse than I originally thought, that it is ALL fake. I am glad you are working on comic books! Thank you.

Expand full comment
Mike Brewster's avatar

Who is doing these is the 64K question. lt would seem to be a fairly limited a fairly small group of people that have this capability and the access to deploy it.

Is it possible to upload these routines after the data base has been created or must they be integral to the design?

Expand full comment
Andrew Paquette, PhD's avatar

Gregg Phillips told me straight to my face who is doing it, and I have had indirect confirmation from others. However, I am not satisfied with these confirmations or Phillips' statement. I do think they are plausible, even probable, but I'd need to see more and have more direct witnesses to be comfortable with any attributions.

Some algos, like NY's Spiral, appear to be static, one-time only uploads. Others, like S1 in WI, appear to be either dynamic, or flexible enough to handle non-S1 compliance with new records.

Expand full comment