What a great prompt.
So here's my hypothesis, from the research I did about MARC history for my
c4l talk list year:
MARC must die, because it was invented before a lot of the tech ideas we
rely on today (notably relational databases) and is therefore solving
problems we don't have and not solving problems we do.
But MARC has *survived* -- not just because it had the phenomenal resources
of the LoC behind it -- but also because the way it was developed was so
empirical and so deeply embedded in the practices of working catalogers. I
don't think it was really even trying to promulgate a standard. I think it
was trying to do exactly what it says on the tin -- to be *cataloging*,
except, you know. Machine readable. This has fossilized practices that
maybe we didn't want it to, but also made it super fit-for-purpose from day
one.
Nowadays there's an awful lot we do with metadata in addition to
cataloging, but of course we still do that too. And I think designing a
replacement with the staying power of MARC doesn't actually start with a
bunch of programmers looking to abstract a framework for metadata; I think
it starts with shoulder-surfing a lot of consumers & creators of
bibliographic metadata and figuring out how they need it to serve them.
Developers ETLing and presenting metadata are one, but only one, of those
constituencies.
On Wed, Oct 22, 2025 at 10:23 PM Roy Tennant <roytennant_at_gmail.com> wrote:
> I know that you haven’t heard from me for years, and now all of a sudden
> I’ve resurfaced. But don’t worry, I only have a few things to say and then
> you can go back to your regular programming, just like I will return to
> fixing trails and rowing rivers. It’s just that you’re the only audience
> who will actually understand what I have to say.
>
> *I have a major confession, an abject apology, and a humble suggestion.*
>
> But first I must set up my key mistake. I will do my best to keep it brief,
> but I’m sorry to those of you who may have heard some of this before. I
> just want to make sure you all have the relevant background to understand
> how I arrived at my key mistake.
>
> I was the lead author with two others for the first book about the Internet
> for librarians (we capitalized “internet” back then), and one of the first
> dozen or so books about the internet for the layperson *total*. I parlayed
> that into a speaking and writing career that took me to a number of
> countries around the world. I wrote or edited half-a-dozen books. I was
> regarded as somewhat of a library technology leader.
>
> And then, toward the end of my career, I blew it. I made the wrong
> decision. Here’s how it happened, and why it matters.
>
> To reduce a lot of years into a small summary, I felt like I was a bit like
> Forrest Gump, where I found myself at the center of so much of the major
> transition that libraries were experiencing from the mid-80s to the
> mid-90s. If you were at a major university you experienced this earlier
> than if you were at a public library, simply because universities were
> plugged into the Internet earlier than public libraries. But public
> libraries totally experienced automating circulation and the card catalog
> as well as the CD-ROM library index revolution. And then, the internet.
>
> In my personal trajectory as a library professional, at some point I’d
> essentially reached a point in library technology where I’d come to
> understand that the biggest obstacle to libraries doing what I felt they
> should be capable of doing was the antiquated MARC standard. Of course that
> led to my oft-cited “MARC Must Die” Library Journal column, as well as
> other takes on that in subsequent years that I won’t bother citing.
>
> So when I felt myself at a professional stalling point I looked around to
> where to go next, and given my take on where I thought librarianship should
> go, I thought I should pick who had the wherewithal to fix our essential
> problem—our library metadata.
>
> That was OCLC. Only OCLC had both the data itself (WorldCat) and the
> computing power (a 50-node compute cluster with gigabytes of RAM and a
> Hadoop platform for parallel processing) to do what I thought needed to be
> done. Using this infrastructure I did a lot of processing of MARC, and
> discovered a lot of serious issues, including how difficult it was to know
> when a URL would actually lead you to the full-text of an item (this is
> essentially impossible in a frightening number of cases).
>
> And I tried. But I didn’t try hard enough, or I couldn’t convince a truly
> corporate culture to actually care to do something that didn’t result in a
> profit. Or I failed to find a way to do the right thing and still make a
> profit. Whatever the reason, I failed. This is my confession.
>
> Also, I just have to say that I went along with the whole “linked data”
> thing, because the “keeper of the flame”—the Library of Congress—believed
> that should be the next thing. And so OCLC was going along and I played
> along as well. But I was not happy with it. I just didn’t feel like I could
> oppose it. Partly I was simply unsure. I didn’t know where this new
> technology might go. Instinctively, I felt like it was too complicated for
> what we needed (frankly I just wanted a much better MARC), but I didn’t
> feel positioned well to oppose it. Another failure.
>
> I was let go at the end of August 2018 because I was not suited to put on
> events. Well, yeah. It wasn't the reason why I joined OCLC to begin with,
> but I take full responsibility for not understanding the job I was
> accepting or had become over the years. It’s on me.
>
> I just wish I could have made a better case for OCLC to do what only OCLC
> could do. To do the metadata processing of MARC that could have really
> brought it truly into the computer age. So this is my abject apology.
>
> I truly wish I had found a way to make a real, lasting difference for
> libraries when I saw it in front of my face. I didn’t. I’m sorry.
>
> Please don’t feel sorry for me, or think you need to reassure me about my
> impact on you or the profession. I’m really not looking for reassurance or
> approval. I just want to set some things straight about my positions on
> library metadata, especially given my association with OCLC and the Library
> of Congress. And I want to publicly own my mistakes, which seems like the
> right thing to do at this point in my life.
>
> If I could end this with one suggestion, it would be for CODE4LIB to take
> on designing a revised metadata standard that was actually well-designed
> for machine processing. Design something that solves problems without being
> a pain in the ass like linked data, and make a real case for it. Set up a
> non-profit organization to manage it and change the library world. If
> there’s any group of people best positioned to do this, it’s you!
>
>
> You have my love and respect, even while you may be thinking about what an
> idiotic suggestion I just made. And all that's OK.
>
> Roy
>
--
Andromeda Yelton
Lead Software Engineer, JSTOR Labs
https://andromedayelton.com
@thatandromeda (Mastodon <https://ohai.social/@thatandromeda>, Bluesky
<https://bsky.app/profile/thatandromeda.bsky.social>, github
<https://github.com/thatandromeda>)
Received on Fri Oct 24 2025 - 19:28:12 EDT