Five Things: September 19, 2024
This is my bi-weekly “newsletter” delivered straight to your inbox, with at least one guaranteed typo I’ll catch after hitting send! If email’s not your thing, don’t hesitate to switch to the RSS feed. I encourage you to click through the main links (there’s only five and they’re all interesting!), and if you enjoy something — send me (or them) an email, leave a comment, or hit the socials!
NOTE: Not directly related to anything in the newsletter, but if you’re on LinkedIn, you might want to turn this setting off, unless you’re okay with them using your data for training content creation AI models and not paying you for it.
_ONE
Hachette v. Internet Archive: We’re Still Fighting for Fair Use | Jennie Rose Halperin
The Second Circuit Court stuck to the facts of the case, which was about IA lending digitized copies of books that were licensable: they did not address every implementation of CDL by libraries.
This is a level-headed take on the Internet Archive’s predictable own goal. The ruling doesn’t jeopardize legitimate libraries’ acquisition and lending of ebooks, nor does it kill other reasonable implementations of Controlled Digital Lending (CDL) outlined here — including what IA was doing before recklessly deciding to ignore its core principles at the start of the pandemic.
As a general rule, I wholeheartedly believe publishers have treated libraries poorly for years when it comes to digital content, thanks to a combination of willful ignorance, misinformation, and an antiquated business model that denies shifting consumer preferences. But there’s no question that IA grossly overreached by ignoring its own 1:1 rule for a couple of months with their National Emergency Library, a change some believe was opportunistically made to force the lawsuit they inexplicably thought they could win.
In the meantime, real libraries [don’t even…] have been pulled into their mess, shamelessly used as goodwill shields to drive support and donations for IA’s self-inflicted wound — occasionally pitting authors and libraries against each other along the way — despite the vast majority of libraries offering access to a wide range of ebooks through legal channels that generate revenue for publishers and authors. Many of those libraries are also actively challenging unfair licensing terms for digital content by engaging publishers and distributors directly, as well as through proposed legislative measures — an effort Library Futures has taken a leading role on.
It’s particularly frustrating because IA is behind one of the most invaluable digital initiatives I can think of (the Wayback Machine), and their hubris potentially threatens that important work.
DISCLOSURE: I’ve recently joined the advisory board for Library Futures, an organization I’ve admired watching methodically find its footing in the very difficult intersection of publishers, libraries, ebooks, and copyright law. I haven’t been directly involved in anything with them yet, but moving forward, if I ever mention LF’s activities, I’ll be sure to note my general advisory role and if I had anything to do with a specific initiative, statement, etc.
__TWO
It is Imperative That We Do Better At Communicating E-Book Pricing to Our Patrons | Becky Spratford
I really want September to not be about tallying how many new library cards you get and instead be about how many conversations you have had with your community about what the library offers. And how we use their money to serve them is a conversation starter. When they are being fleeced for the things they want, when they know the truth, they will get mad and advocate right alongside us. Once that happens, our legislators will take action.
Speaking of real libraries and legislative options, the more librarians start taking matters into their own hands, the better — and Spratford nails the call-to-action here. 10/10, no notes; just read it.
Related: Shots fired in Newburgh!
___THREE
Orange Inc exec clarifies AI use for emaqi manga localization | Deb Aoki
One thing that struck me in the press release for emaqi that went out this week was that this time, you’re a lot less vague about how much you’re using AI for manga localization.
There’s a great mix of questions here, including pointed follow-ups, about a $20m manga startup that made a big deal about AI-driven translation — and is now trying to walk that back a bit. Aoki has deep expertise in manga and UX design, and reading this was like watching an ace pitcher bedazzling hitters, recording strikeout after strikeout on the way to a no-hitter.
I’m kind of surprised there wasn’t a PR flack standing by to shoo her away every time she unleashed some high heat or a wicked curveball after setting them up with a few purposeful soft balls! This is how all journalists should handle covering startups (or any companies) that are making bold claims about their products, especially right now with anything AI-related.
As for emaqi, I wouldn’t be surprised to see them pivot to being a service provider, selling AI-powered translation services to third-tier publishers who will flood the market with translations of little-known titles there’s no audience for beyond their native languages or the bottomless wells of digital platforms chasing quantity over quality. More likely, though, it’ll just be another $20m lost to AI hallucinations, most notable for getting a rare bit of critical media coverage.
____FOUR
The LLM honeymoon phase is about to end | Baldur Bjarnason
The usefulness of LLMs was always overblown, but unless the AI vendors discover a new kind of maths to fix the problem, they’re about to have an AltaVista moment.
This is an insightful read, spotlighting a noted tech journalist discovering the internet hates him, and how he tries to “fix” the internet.
One of the biggest flaws with tech journalists and sycophants is their collective willingness to ignore the past and avoid connecting useful dots to the present. Bjarnason, a discerning practitioner, is always good for it, and by exhuming AltaVista’s corpse, comparing the inevitable manipulation of LLMs to old school black hat SEO, he delivers a no-brainer outcome not enough people seem to have considered.
Of course, while it’s more fun to pretend we might be heading towards Skynet, the reality is much less exciting — but equally threatening. “AI” is just another symptom; the people using it to advance their nefarious schemes are the real problem.
_____FIVE
AI and the American Smile | jenka
Every American knows to say “cheese” when taking a photo, and, therefore, so does the AI when generating new images based on the pattern established by previous ones. But it wasn’t always like this.
Apologies for the recent uptick in AI-related stuff, but this one comes at it from a different and fascinating angle that I appreciated on a couple of levels.
First, it’s a great example of the way inherent biases and limited datasets warp the output of these systems. Equally important, though, as someone who has always hated being forced to smile in pictures, it’s validation that I’m not wrong!
If you’re reading this somewhere other than your inbox and would like to get it via email, sign up here, or you can add the RSS feed to your favorite reader. You do you!
Discover more from As in guillotine...
Subscribe to get the latest posts sent to your email.
::like::
On the Internat Archive: they had earned a LOT of collective good will for several reasons, the Wayback machine being one of the most obvious, and squandered in ways that no one can explain. It’s just…beyond baffling.
On Library Futures: good going! And best of luck.
On the AI bubble: the damage this bubble is doing to the environment is beyond the pale–especially that the press is not shaming all of them into oblivion. But then, who owns the press but the same class of people who is so insulated by wealth they truly believe climate change won’t touch them in any way.
I feel like the tide is very slowly turning on lesser forms of AI, at least, but I think we’re in for a long “pivot to video” style overreaction by the big tech companies, and ripple effects from the inevitable course correction that’ll be required is going to be 10x more disruptive. 🙁