Emma Stanford

Screenshot of suggested correction of ORCID iD to ORCID ID

Usually Google Docs can shut up with its suggestions but this time I couldn’t agree more, ORCID

The coolest little #IIIF image tile demo from @latest.allmaps.org: observablehq.com/@allmaps/til... (I zoomed in on the dog because of #branding)

Every time Taylor Swift sings "And the tennis court was covered up with some tentlike thing" I think to myself "It's called the Lung, Taylor" and I'm sharing this in case there are any other Swifties out there who have a complicated relationship with Infinite Jest

(Question prompted by today's very cool session at the David Rumsey Map Center)

Are people really using ChatGPT to answer cataloging questions (e.g., "When was X railroad built so I can date this map")?

Diplomatically: it seems unwise to attribute informational value to ChatGPT responses when we know it can hallucinate and we can't control what sources it uses. 📜

Screenshot of LOC search results for "Books", with 600,000 results

Just now found myself typing "books" into the Library of Congress linked data service and hitting "search" and I'm worried that if I don't get this out in the open right away someone will try to blackmail me with it

Found one, although obviously what I'm really looking for is a high-res image to put on my wall royalsocietypublishing.org/doi/10.1098/...

Hot damn this is a good blog post. Looking for a digitized copy of that de la Rue chromolithograph led me to this article, which reproduces some of the photographs themselves: artsandculture.google.com/story/starin...

(This is the medium model and I think it's safe to say my standard-issue 2020 MacBook Pro can't handle it; small worked a lot better)

Screenshot of Whisper transcript including repeated line "I am a man of many sorrows"

It's also picking some incredibly poignant lines to repeat

Screenshot of in-progress Whisper transcription in a terminal, showing a phrase repeated 10 times

Screenshot of in-progress Whisper transcription in a terminal, showing a phrase repeated 10 times

I know OpenAI is evil but I'm running Whisper on my laptop and guys it's just a little buddy trying its best

I'm very optimistic about my profession, I don't know what you mean

Very yellow-tinted images of pages of Bodleian Library MS. Auct. F. 4

I'm teaching some students about digitization and thought it would be fun to estimate how many years it will take to digitize all our collections at our current rate. Apropos of nothing, this manuscript is 1200 years old and these images are 30 years old. digital.bodleian.ox.ac.uk/objects/783d...

That's a great point. I think a similar experiment now might do better (researchers have taken more images and have more experience tagging--what's that app's name?) but in the current climate we'd probably be told to use AI instead of paying staff. Which might work a little, but not a lot.

I agree. I hope I didn't misrepresent Andrew's argument, but I think the risks of relying on vendors and external partnerships for infrastructure are pretty clear. We need more shared tools for building our own infrastructure and the money to pay in-house experts.

I'm inclined to agree. I took Andrew's point as more "If public institutions aren't given enough funding to build sustainable infrastructure, they'll have to find the money somewhere else, and this is one least-worst way." But I hope the BL attack inspires more public funding of infrastructure.

Incredibly foolish to suppose Taylor Swift doesn't know how apostrophes work, guys

(I am very excited for TTPD; I hope it's a mix of maroon/Vigilante Shit/my tears ricochet/illicit affairs; I can't wait to listen to it and pretend I too am eviscerating the small British men who've done me wrong)

Thank YOU Laura! A wonderful session expertly moderated!

Ben Albritton discuss the advantage of a multi-institution model for conservation of digital repositories. 'Changing the one-institution model is one key to sustainability and risk mitigation' @blalbritton.bsky.social

#NYSINYD

Andrew Prescott: 'When you go somewhere like the National Archives, everybody is just taking pictures all day long. You realise that there's a potential there to build a very large resource. [...] But one needs to be able to do it as simply as loading images.'

#NYSINYD

Hi Claire! No problem!

(This sounds amazing but my institution specializes in sensitive 20th- and 21st-century material, so we'd probably have to spend a lot of time chasing violations of our new one-line reading room use agreement, "I WILL NOT POST MY IMAGES ON MANUSCRIPTR")

Andrew asks: Have we thought enough about digital disaster management plans? Something more nuanced than "let's shut everything down right away"? Laura says it's not just disaster management, but the management of inevitable digital change.

Folks are ideating a specialist Flickr/Wikimedia-like tool that facilitates easy tagging and description (integrated with institutions' online catalogues?), allows you to search across your own and others' research images, and isn't going to die or become very expensive in a few years' time.

Btw the Bodleian Flickr project is still available to access: www.flickr.com/groups/bodsp...

Also, on the note of facilitating image downloads: Stewart notes that the BL did allow bulk downloads of their Hebrew MSS , which was unusual and great. (Stewart also mentions his old viral(!) tweet about how to download MS images from various websites. It's still not easy.)

Some questions about how best to support distributed archives of researchers' own images: Elaine mentions the old Bodleian project by Dan Wakelin and Judith Siefring to aggregate researcher images on Flickr.

Panelists note that BL has (temporarily?) prohibited photography in the special collections reading room presumably because CCTV is out and the staff infrastructure for monitoring readers has taken a hit.

We don't want to duplicate the nightmare of bundled e-journal subscriptions, but things like EEBO and 18c Collections Online might be reasonable models for subscription-based digitized resources, says Andrew. (Also plugs BL membership as a way to support!)

Andrew Prescott notes many knock-on effects of BL crisis. British legal deposit service and UK Web Archive both on pause. Digitized MSS perhaps less of a priority atm. BL's funding for digital access is limited--do we need to look to commercial partnerships for dependable infrastructure?