Michael Roberts is categorically not building machetebots
vivtek.com
Michael Roberts is categorically not building machetebots
@vivtek.com
Translator, code monkey, aspiring roboticist, coffee grower, roof builder (emeritus), AI archaeologist, machete wielder, lapsed Quaker, vecino Adjunteño desde 2017 pq Ponce es Ponce y tiene el Sams Club pero el campo es el campo
"Signs point to yes"
November 15, 2025 at 4:00 AM
Kind of the opposite of old girlfriends, really.
November 15, 2025 at 3:59 AM
I'm thinking you don't want equipment getting hot that gets inserted into your skull. Call me crazy.
November 15, 2025 at 3:48 AM
This argument will definitely make Martha's Vineyarders love him enough to sell him a damn pierogi.
November 15, 2025 at 3:47 AM
I think I'm gonna watch it again this week.
November 15, 2025 at 3:42 AM
Oh man, we saw Speed Racer COMPLETELY at random as the second movie in the same night and I don't even remember what the first one was because Orion and I (he was about 10 at the time) LOVED IT TO PIECES and Agi had no idea what she'd just seen.
November 15, 2025 at 3:42 AM
Also this, wondering whether Wells Fargo, which had been foreclosing on the building, didn't have something to do with it.
therealdeal.com/chicago/2025...
Inside real estate fight that led to South Shore immigration raid
City of Chicago inspected the troubled 130-unit apartment complex two weeks before massive federal operation on Chicago’s South Side.
therealdeal.com
November 15, 2025 at 3:36 AM
Hi, Oblivion! How's the wife and kids?
November 14, 2025 at 7:07 PM
This whole thread is killing me. Thank God I don't even have to read the article.
November 14, 2025 at 7:03 PM
I do a lot of poking around in papers that are (for the AI field) really old, so scan quality and font kind of remind me roughly when something was published and sometimes in which journal. When all I can find is somebody's retyped Word document or whatever I can barely even manage to deal with it.
November 14, 2025 at 6:59 PM
And the *reason* I don't want to just OCR it and then run it through Typst or something is that the memory of the page layout and font is pretty important in how my brain indexes things. I don't know how universal that part is.
November 14, 2025 at 2:35 AM
The scanned image and the text layer for each page are separate objects in every PDF. Most of the time, when you download a PDF from somewhere, it's already got that text layer - but without it, you can't select (or highlight) text on the page. I've got a few I want to reverse-engineer.
November 14, 2025 at 2:33 AM
Hey, I have a bunch of researchers here, too. But also I'm trying to get into the habit of cross-posting the important stuff. Which this *is*.

I should probably move cat pictures into the important stuff category, too. You can't go wrong with cat pictures.
November 14, 2025 at 1:32 AM
Answer (from the Fediverse, so you need to up your game, Bluesky): OCRmyPDF.

github.com/ocrmypdf/OCR...
GitHub - ocrmypdf/OCRmyPDF: OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched
OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched - ocrmypdf/OCRmyPDF
github.com
November 14, 2025 at 1:21 AM
As a member of the Hoosier Diaspora I *hate* seeing Indiana in the news about 98% of the time.
November 14, 2025 at 12:45 AM
THE WORST
November 13, 2025 at 10:58 PM
Personally, I think he should be dropped at the side of a road in rural Chiapas without wallet, ID, or shoes.
November 13, 2025 at 10:23 PM
Just weird they scream law and order and then brag about their intent to break the law and create disorder.
November 13, 2025 at 10:20 PM
I've always said they're perfectly fine with grooming - they just can't stand the idea of competition.
November 13, 2025 at 10:17 PM
Nobody ever studies the literature any more. I blame Rich Sutton.
November 13, 2025 at 10:10 PM