Journal Entry
I spent the afternoon calibrating and testing the computers as best I could without the Commander's unique abilities. I requisitioned two astromechs for the day and put them to work using the surveillance cams to follow me, and other people I had them select based on set criteria, around the halls of the complex.
Of course, I limited the test to the corridors, lifts, and other spaces that should be free of sensitive information. The droids then went on a scavenger hunt through the samples, routing their finds to the display screens on the walls of the office, through the recording and download process, or out to the data pad and headset I had on me.
The search criteria I gave them were all mundane: specific types of droids, specific innocuous scenarios. Then I put the cataloging system through its paces in test mode, to get the hang of its differences from the civilian systems I'm used to.
All this testing was well worth it. We caught and fixed a glitch preventing sound from the feeds from passing to the cataloging system. Everything else checked out, and that was the last item on my list, so I think all is as ready for the Commander's return as I can make it.
[Sample of testing data - not included in the journal entry]
Len: ~200 words
Criteria met: Two or more people having a conversation in a lift.
Subjects: Two human aids in lab uniforms who had, coincidentally, just passed Blithe in the halls.
(Continue reading for transcript.)
[End sample]
S1: Was that another probie? Did you get a look at her insignia?
S2: Think so, yeah. There's more of them around, lately. You think she's from that lab in the, uh... rumors - you know the one?
S1: The one? Oh! Oh, yeah. I hear the clearance rating's so high you'd have to be a moff to even know where it is. Nah. She's not from there.
S2: No?
S1: Didn't look the part.
Subjects exit the lift and make their way to the nearest break room.
S2: How so?
Subject 1 checks around them before continuing.
S1: I hear they're… different.
S2: Different like how? Non-humans? That's not --
Subject 1 interrupts, voice lowered but still audible.
S1: Like not military intel or ISB, or anything anyone recognizes. I hear they've got custom, like really custom uniforms, and super fanatical faces. Real intense. And I don't know this, but I also heard the assistants never, ever leave that lab. At all.
S2: What do --
Subject 2 lowers her voice, too.
S2: What do you mean never?
S1: They say they go in and…
Subject 2 interrupts by clearing her throat as a technician enters.
S2: Say, does this caf taste funny to you?
6 notes
·
View notes
*Meaning: When did you stop wearing a mask to a majority of your public activities? Wearing a mask when you feel sick or very rarely for specific events/reasons counts as “stopping”
*Edit: This post is about COVID not Autism*
Edit 2: Please vote in updated poll Here with clearer wording, thank you!
224 notes
·
View notes
look computational psychiatry is a concept with a certain amount of cursed energy trailing behind it, but I'm really getting my ass chapped about a fundamental flaw in large scale data analysis that I've been complaining about for years. Here's what's bugging me:
When you're trying to understand a system as complex as behavioral tendencies, you cannot substitute large amounts of "low quality" data (data correlating more weakly with a trait of interest, say, or data that only measures one of several potential interacting factors that combine to create outcomes) for "high quality" data that inquiries more deeply about the system.
The reason for that is this: when we're trying to analyze data as scientists, we leave things we're not directly interrogating as randomized as possible on the assumption that either there is no main effect of those things on our data, or that balancing and randomizing those things will drown out whatever those effects are.
But the problem is this: sometimes there are not only strong effects in the data you haven't considered, but also they correlate: either with one of the main effects you do know about, or simply with one another.
This means that there is structure in your data. And you can't see it, which means that you can't account for it. Which means whatever your findings are, they won't generalize the moment you switch to a new population structured differently. Worse, you are incredibly vulnerable to sampling bias because the moment your sample fails to reflect the structure of the population you're up shit creek without a paddle. Twin studies are notoriously prone to this because white and middle to upper class twins are vastly more likely to be identified and recruited for them, because those are the people who respond to study queries and are easy to get hold of. GWAS data, also extremely prone to this issue. Anything you train machine learning datasets like ChatGPT on, where you're compiling unbelievably big datasets to try to "train out" the noise.
These approaches presuppose that sampling depth is enough to "drown out" any other conflicting main effects or interactions. What it actually typically does is obscure the impact of meaningful causative agents (hidden behind conflicting correlation factors you can't control for) and overstate the value of whatever significant main effects do manage to survive and fall out, even if they explain a pitiably small proportion of the variation in the population.
It's a natural response to the wondrous power afforded by modern advances in computing, but it's not a great way to understand a complex natural world.
126 notes
·
View notes
This is important to me, I need this for data collection and evidence/reference material.
If you are a fnaf fan, be honest, would you pay $30 for a Youtooz DJ Music Man figure if they made one?
It’s been nearly 2 years and they still haven’t made any official DJMM mech besides those playing cards featuring DJ as one of the characters and MatPat’s Fnaf collection in Theory Wear including 2 DJ music Man t-shirts. That’s it. (That I am aware of) There are no official figures or plushies at all. Not from Youtooz or Funko. Meanwhile all the other SB characters including the WET FLOOR SIGNS have official merch..
So- I ask you this question, please answer as honestly as possible:
Please respond whether or not you are a fan of this character
If successful, I am going to use the data in this poll as evidence for an email to Youtooz, so if you can spread this poll as much as possible for a big sample size outside of just the DJ Music Man fan community it would be super helpful! Thank you! And please answer honestly!
Including pictures for anyone who may not know what character is being referred to:
124 notes
·
View notes
trying to work through a thought on vampire/human relationships but then i remember that the entire de romanus bloodline are outliers even amongst their own kind and therefore should not be considered representative of the population
58 notes
·
View notes