The Problem with Library Tech Content
Most library tech writing is vendor marketing dressed up as advice. You've seen it: claims with no sources, case studies that skip the part where everything broke, "best practices" that just happen to be what the vendor sells. It's a genre.
Good research exists, but it's scattered across academic journals, government reports, practitioner blogs, and nonprofit studies. And almost nobody publishing in this space tells you how they verify what they write, or what they don't know. So here's that page.
How I Source Information
Primary Sources
I read the actual documents whenever possible. Full text of legislation (SB 24-205, EU AI Act, state privacy laws), actual lawsuit filings instead of news summaries, real contract language and SLAs, government audits and regulatory guidance. If someone cites a statistic, I go find the original study. If a vendor makes a claim, I check the contract.
Research and Reporting
I pull from library science journals, computer science papers on security and AI, policy analysis from tech-focused think tanks, and practitioner writing (blog posts, conference talks, case studies from librarians doing the work). I also lean on IMLS data, BLS statistics, FTC guidance, ALA reports, and OCLC research. When I cite Gartner or Forrester, it's for analysis, not the hype cycle graphic.
What I Don't Use
Vendor marketing materials get read for critique, not citation. I don't use unverified social media claims, AI-generated text as a source, or secondary reporting without checking what it's actually citing. If I can't trace a claim back to something real, it doesn't go in.
Geographic Limitations
I focus on US law and the EU AI Act because those are what affect US library vendors. Coverage is strong for federal law and major states (CA, NY, TX, IL, CO), adequate for the EU AI Act and UK GDPR, and weak to nonexistent for Australia, India, ASEAN, and most other regions. That's a real gap and I don't pretend otherwise.
How I Verify Claims
The short version: I don't take anyone's word for anything, including my own.
When someone says "67% of libraries are adopting AI," I find the original study, check who was surveyed and how many, look at the margin of error, compare it to other data, and check whether this is 2024 data or something from 2019 being recycled. A lot of library stats are zombie numbers that won't die.
I read the actual statute, not the press release. For something like Colorado's SB 24-205, that means the full text, any regulatory guidance, and conversations with librarians who've actually tried to implement the requirement. Laws are vague more often than not, and I try to say so instead of pretending there's a clean answer.
"99.9% uptime" is marketing until you check the SLA in the actual contract, pull up the public status page history, and ask customers off the record whether they've actually hit that number. Vendors are consistently vague in marketing and consistently specific in contract exclusions. I look at both.
If I recommend something, I've either tested it myself or talked to practitioners who have. I try to be clear about the difference between "I've done this" and "I've researched this." Implementation case studies matter more than feature lists. Trade-offs matter more than endorsements.
Where I'm Biased
Everybody has biases. Most people writing about library tech just don't tell you what theirs are. Here are mine:
I'm a consultant
Recommending consulting could benefit me financially. In practice, most of what I publish recommends free tools and vendor-independent approaches. But you should know the incentive exists.
I'm a former vendor insider
OverDrive (#70 employee), Baker & Taylor, collectionHQ, Trellis Law. I know how vendors think because I've been in the room. That gives me useful context. It also means I see vendor problems clearly, maybe more clearly than the other stuff.
Public libraries and consortia are my wheelhouse
Academic library coverage comes from research, not lived experience. Course reserves, research data management, faculty politics -- I know these exist but I haven't navigated them personally.
US-centric with an EU exception
I track US law and the EU AI Act. Everything else gets much less attention. That's where my expertise is, but it's still a limitation.
Large systems are easier to see
Big library systems have documented failures and clear metrics. Small libraries might have solutions I don't know about because nobody wrote them up. Scale bias is real and I'm subject to it.
I chase what's urgent
New legislation, breaches, AI hype cycles -- that's what I write about. Evergreen library management topics are underrepresented here because I'm drawn to what's on fire.
Update Cadence
Major legislation and breaches get published as I learn about them. Research articles get a quarterly review for new data. Legal analysis (AI Act, state privacy laws) gets an annual refresh, usually in January. Corrections happen immediately. I don't sit on known errors.
How I Handle Errors
When I get something wrong, the correction goes at the top of the article with the updated date, the original claim stays visible so you can see what changed, and I explain why it was wrong. No stealth edits. If you find an error, tell me and I'll fix it within a week.
What I Don't Know
There are real gaps in what this site covers. I'd rather tell you than have you find out the hard way:
Academic libraries
I know the public library and consortia side. Academic-specific challenges are covered from research, not from having done the work.
International law beyond the EU AI Act
Canada, Australia, UK privacy laws -- covered lightly at best. If you need international legal analysis, this isn't the site for it.
Every vendor platform
I haven't personally implemented every system I write about. I research extensively, but there's a difference between research knowledge and hands-on knowledge, and I try to flag which one I'm working from.
Open-source long-term viability
Some of the projects I cover are relatively new. Whether they'll still have active communities in five years is genuinely uncertain.
How I Use AI (And How I Don't)
I use AI tools to help draft, structure, and edit. The research is mine. The conclusions are mine. The AI doesn't decide what's true or what gets published -- I do.
I do not use AI to invent sources, fabricate quotes, or generate analysis. Every claim still goes through the same process: primary sources, verified numbers, cross-referenced against multiple references. When the model and the evidence disagree, the evidence wins. That happens more often than you'd think.
I don't paste patron data, contracts, or anything confidential into AI systems. Ever. And I pay for tools that are explicit about data handling, because if I'm going to yell at vendors about data practices, I should probably not be feeding library data into a black box myself.
How to Audit This Work
Don't take my word for it. I mean that literally. Click the sources. If I cite a study, go read it. Check whether the data is from 2024 or 2019 -- a lot of library tech stats are zombies that won't die. See if other people agree or disagree. Notice when I'd benefit from what I'm recommending. And if you catch me being wrong, tell me. I'd rather be corrected than confidently wrong.
A Note on Vendors
Not every vendor is the enemy. Some vendors do good work, charge fair prices, and actually pick up the phone when something breaks. This site focuses on the patterns that don't work because those are the ones nobody else is calling out. If your vendor relationship is healthy, great. Keep them honest, keep the contract fair, and keep your options open. That's all I'm saying.
Found an Error? Different Perspective?
Contact me -- I take corrections seriously and will update immediately. Disagreements about interpretation are welcome too.