[dperfect]

Nerdsnipe confirmed :)

Claude Opus came up with this script:

https://pastebin.com/ntE50PkZ

It produces a somewhat-readable PDF (first page at least) with this text output:

https://pastebin.com/SADsJZHd

(I used the cleaned output at https://pastebin.com/UXRAJdKJ mentioned in a comment by Joe on the blog page)

[pests]

So it was a public event attended by 450 people:

https://www.mountsinai.org/about/newsroom/2012/dubin-breast-...

https://www.businessinsider.com/dubin-breast-center-benefit-...

Even names match up, but oddly the date is different.

[elmomle]

Your links are for the inaugural (first) ball in December 2011; OP's text referred to a second annual ball in December 2012.

[pests]

You are right my first is incorrect but the second does seem to be from 2012.

[sorbus-25]

DUBIN BREAST CENTER SECOND ANNUAL BENEFIT MONDAY, DECEMBER 10, 2012 HONORING ELISA PORT, MD, FACS AND THE RUTTENBERG FAMILY HOST CYNTHIA MCFADDEN SPECIAL MUSICAL PERFORMANCES CAROLINE JONES, K'NAAN, HALEY REINHART, THALIA, EMILY WARREN MANDARIN ORIENTAL 7:00PM COCKTAILS LOBBY LOUNGE 8:00PM DINNER AND ENTERTAINMENT MANDARIN BALLROOM FESTIVE ATTIRE

[Groxx]

Since it looks like this got flagged (probably because out of context at a glance it looks like insane babble that somewhat frequently occurs here), some context: this is appears to be text recovered from the pdf, in the links up-thread. Though there's more text than that link shows, and I'm not entirely sure why it's posted in this specific thread, though it's relevant-ish at least.

[sorbus-25]

It's from a contemporaneous reference to the very same event listed in the PDF. I found it online and archived it: https://web.archive.org/web/20260206040716/https://what2wear...

It includes screenshots of what looks like an expanded document for the event.

Why relevant? I found it by searching the archive for "DBC". There were references to "Dubin", then I found the rest online easily. All that extra text could have helped with decoding the base64 text

[turtlesdown11]

interesting, Eva Dubin was highlighted today for offering Epstein her 15 year old daughter and her friends.

She's a medical doctor, who became amnesic when on the stand for Maxwell's case

>Pressed about gaps in her memory, Dubin told the court: "It's very hard for me to remember anything far back and sometimes I can't remember things from last month. My family notices it. I notice it."

[nialv7]

looks like we have it. in the end it's pretty mundane...

[JKCalhoun]

There are plenty of other PDF's with Base64 encoded attachments.

[klustregrif]

Which begs the question why was it censored?

[nickthegreek]

They censored Dont in one location. The current Thought process is they were redacting mentions of Don T.

[pbhjpbhj]

Mighty be they censored all pages mentioning keywords, this one says "Breast" ... perhaps they censored all sexual content?

[klustregrif]

At the risk of repeating myself. Which begs the question why?

[mmastrac]

To protect the people in power, as always.

[mikeyouse]

Likely because the named list is a bunch of Trump appointees and mega donors and they're illegally trying to spare them the embarrassment.

[lanyard-textile]

Distraction.

[notpushkin]

> It produces a somewhat-readable PDF (first page at least) with this text output

Any chance you could share a screenshot / re-export it as a (normalized) PDF? I’m curious about what’s in there, but all of my readers refuse to open it.

[dperfect]

Screenshot: https://imgur.com/eWCfYYd

[dperfect]

Letting Claude work a little longer produced this behemoth of a script (which is supposed to be somewhat universal in correcting similar OCR'd PDFs - not yet tested on any others though): https://pastebin.com/PsaFhSP1

which uses this Rust zlib stream fixer: https://pastebin.com/iy69HWXC

and gives the best output I've seen it produce: https://imgur.com/itYWblh

This is using the same OCR'd text posted by commenter Joe.

[daveguy]

> which is supposed to be somewhat universal in correcting similar OCR'd PDFs

Xerox would like a word.

https://news.ycombinator.com/item?id=29223815

Point being, "correcting" to "correct looking" may be worse than just accepting errors. Errors are often clearly identified by humans as a nonsense word. "Correcting" OCR can result in plausible, but wrong results that are more difficult for the human in the loop to identify.

[dperfect]

That's true if we're correcting OCR of actual output text. In this case, it's operating on the base 64 text, trying to produce chunks that form valid zlib streams and PDF syntax so the file can be intact enough to be opened. "Just accepting errors" would mean not seeing any content in the file because it cannot be read.

So yes, the "fixed" output has errors, but it’s not hallucinating details like an LLM, nor is it trying to produce output that conforms to any linguistic or stylistic heuristics.

The phrase "correcting similar OCR'd PDFs" should have been "correcting similar OCR'd base 64 representations of PDFs".

[the_real_cher]

This is cool!