Why we need journals (dammit)

10 Nov 2009

The always interesting Jon Katz has recently posted a number of thought-provoking articles on the state of computer-security research and some problems therein. (See here and the follow-up here.) I have my own thoughts on the general issue (which I will post about later) but wanted to quickly reply to one particular suggestion from the comments: that we don't need journals, or at least we don't need paper journals.

In reply, I would like to state as clearly and as emphatically as I can: we need paper journals. Why? Because we are not the end of history.

There will be future generations of researchers, they will (we hope) want to build on our work, and so they will need to be able to find our work. We have a responsibility to future generations of researchers (and those who would benefit from such research, which is everyone) to make our research as available in fifty years and a hundred years and two hundred years as it is today. We have an obligation not only to disseminate our papers but also to archive them in a stable and durable way-- and that means paper journals.

Don't believe me? Let me tell you a little story. I have been honored to serve the Computer Security Foundations Workshop/Symposium as Publications Chair for the past eight years or so. (I honestly lose count.) On one of those years (2002) we celebrated our fifteenth anniversary and wanted to mark the occasion by issuing a CD with electronic copies of all papers from all fifteen of those years. Fortunately, the previous Publications Chair (Joshua Guttman) had kept an electronic archive of papers from year eight (1995) until I took over. So, we only needed to collect electronic copies of years one (1988) through seven (1994). And as Publication Chair, it was my responsibility to contact all of the relevant authors to request any electronic copies they might have. I forget the exact figures (or even the rough ones) but this entailed contacting roughly 150 to 200 authors about 100 to 125 papers. Wanna guess how many electronic copies I received?

Less then five.

Instead, I received excuse after excuse like the following:

  • "I'm sorry, but I can't find the files any more."
  • "I'm sorry, but it seems those files were improperly restored after a drive-crash two years ago."
  • "I'm sorry, but I changed jobs and don't have access to those files any more."
  • "I'm sorry, but I don't have a device that can read the relevant disks any more."
  • "I'm sorry, but all the relevant computers burned up years ago." (Really. I got this one).

Effectively, more than 95 percent of these papers had ceased to ceased to exist in electronic form. We had to scan in the papers from paper copies of the proceedings.

The big lesson I extracted from this is that we do not know how to archive papers in electronic format yet. We do know how to archive papers, though: we print them out on paper (acid-free paper, no less) and put copies in as many libraries in as many universities in as many countries on as many continents as we can. We know how to archive paper. We've been doing it for centuries. Electronic formats? Not so much.

The situation may have improved since 2002, sure. The PDF format seems fairly stable, and there are lots of free readers for it. And there are good reasons to augment paper journals with on-line databases that allow you to search based on meta-data. But that's not the issue-- the issue is whether we still need paper journals. And if that's the question, the answer is 'yes'. We have an obligation to archive our work using the most reliable, well-understood and poven technology we know of, and right now, that's paper.

(If you need further convincing, I issue you a little challenge. Go to your university library and find the oldest journal they actually still have on the shelf. Now, go into your own archives and find the oldest non-ASCII printable document (Word, TeX, nroff, etc.) that you can print right now, on your current computer system, without translation or updating. Is it actually the oldest such file in there?)

Share this

We may not disagree...

We may not disagree. (I love books/newspapers/magazines. Then again, I am over 35.) It could be that we end up with a system where journals are published both electronically (these will be the ones that people will actually use) as well as on paper (for the Library of Congress and, say, Harvard). But the economics of having more than a handful of paper copies simply doesn't make sense -- no one reads them, and libraries increasingly can't afford them.


Printing costs << total purchase costs

Andromeda said it before I did, but I also have the impression that the printing and shipping costs are actually a small part of the purchase cost of a journal. But I'd go further: based on my experiences with CSFW proceedings, the big costs of a journal issue or conference proceedings are fixed, not marginal. Therefore, if you want to lower the purchase cost of a single copy, you'd want to print as many of them as you can. In that way, you spread the fixed costs over many individual copies, lowering the per-issue purchase cost. Eventually, yes, you'd reach the point where per-unit cost equals marginal costs, but we're not there yet. And I don't think demand will ever rise to the point where we will be.

(This discussion is just about purchase cost, though. There is also the ongoing costs of storage-- but there are subscription fees for on-line journals too.)


There are initiatives out

There are initiatives out there to provide for archiving of ejournals, although I don't remember any of the specific initiatives or technology, it being 7:30am and pre-coffee (the best I can do is wave the magic wand of cloud computing). Ownership and persistence of ejournals are problems libraries are pretty concerned with -- with a paper journal, you know you own it and you know format changes aren't a problem (even though there are long-term archival problems for paper as well), but with an ejournal...are you renting the content from the database aggregator? What happens if they go belly-up? Libraries would like to have permanent copies of this stuff, not just the right to licensed access, but that's not what's implied by the structure of things these days, hence the attempt to produce neutral repositories.


Better format for archiving?

I'm sure that there are long-term archival problem with paper as well-- heck, some of my paperbacks fall apart after only ten years. But my point is that by now, we understand what those problems are and how to mitigate them. I have no such faith for things like hard drives, CDs, DVDs, etc. And that's not even considering the problem of making an electronic format (PDF, TeX, Word, etc.) readable for decades. (I'd believe that ASCII would be readable over that long, but what about formulae, pictures, diagrams, graphs, and so on?)

But let me ask you, my librarian friend: if journals are going to bifurcate into a convenient/accessible form and a long-term archival form, what would *you* choose for the second? Paper? Microfiche? Laser-etched steel plates?


Obviously anything with

Obviously anything with lasers is awesome ;), but preservation is one of the classes that didn't quite make it into my list, so I'd just be talking out my ass there; I'd want to consult with someone who actually knew the issues. I guess I'd want something durable, with manageable storage costs, that did not require much in the way of specialized equipment to read (or, at least, whose specialized equipment is also durable with low storage costs), but I don't know what the best material type choice would be. (May depend on the institution and its archival budget and facilities...)


Jonathan -- Libraries

Jonathan --

Libraries increasingly can't afford ejournals, either; their rate of cost increase is stellar. (And many are published both electronically and on paper these days.) When I think of the disads of paper I don't think cost as quickly as I think of unpopularity with users, and shelf space (which represents cost as well, but indirectly).


Cost breakdown?

Do you have any idea what a journal's cost-breakdown looks like? How much goes to editors, how much to printing and shipping, and so on?


I don't, I'm afraid. I

I don't, I'm afraid. I imagine it depends on the journal. (I wouldn't assume that the lack of printing & shipping costs for ejournals necessarily lowers their total cost, though, as licensing issues become complex.)