Least Recently Used (LRU)

As I was following through on a couple of questions related to gone-noting this week, to my surprise I ended up looking at the Harris & Harris textbook, Digital Design and Computer Architecture (2021), chapter 8, “Memory Systems,” and even more closely, section 8.4.5, “Replacement Policies,” where I found a description of how programmers and computer engineers might name a vanishing edge of retrievable memory. Least Recently Used (LRU) policies provide a processing system with a cut-off, a threshold past which, well, the thing in question gets cached (i.e., reclassified from available to gone). Although less spiritually awe inspiring, it’s sort of like a sky burial for digital objects, where the algorithmic replacement policy is the buzzard and least recently used artifacts are ferried to the great elsewhere-beyond. I don’t anticipate clinging too tightly to computer architecture analogies for the research project I am working on—puzzling through, drafting—but LRUs have given me a lot to think about.

LRU analogs are everywhere. The obvious and most intuitive examples come from computer environments, such as when an iPhone autodiscards a too-long-unused apps. Suppose I last used the Zoom app on my iPhone two years ago; a least recently used policy “evicts” the unused app, frees the memory, and keeps the system fresh, nimble, less crowded and cluttered. It lightens the processing load, a modest sacrifice to efficiency gods. According to Harris & Harris, LRUs rely on a “principle of temporal locality,” which posits that, as in this iPhone example, the app most recently used is likelier to be used again sooner than an app gathering (hypothetical) dust or lagging forgettable toward oblivion. The LRU is gone-going: forgettable, archaic, disposable. Temporal locality reminds me of Jenny Odell’s books, especially where she writes about all we perceive to be accelerating, bustling, and compressing, which oftentimes leads us as mere mortals to feeling fogged, tired, and overextended shells of our best selves. In How to Do Nothing (2019), a chapter titled “Uselessness as Survival,” Odell writes about “Old Survivor,” the only old-growth Redwood in Oakland whose persistence owes to being unsuitable for logging. By extension, old growth forests, rarer and rarer as they are these days, hang back in the arena of least-recentlies, Least Recently Logged (LRL).

In the front shed, somewhere down deep among the tools lurks a Least Recently Purposed (LRP) jimmy jammy. Racked in the mud room are few pairs of shoes, and one of them is the Least Recently Worn (LRW). The refrigerator holds condiments Least Recently Sandwiched (LRS). The pets agitate and jockey for couch positions when they realize themselves to be Least Recently Petted (LRP). Somewhere an overzealous homeowners association president issues a fine to the neighbor whose lawn was Least Recently Mowed (LRM). You get the picture. Temporal locality splits the rhythms of everyday life until there is no slowness, only acceleration. Recentliness, a function of efficiency drive, acts as a winged chariot and speeds us, time doing its flying, us to the grave.

Figure 1. Cover of the inaugural issue of the Journal of Advanced Composition, published in Spring 1980.

But LRUs also connect with questions of disciplinary epistemology and memory: What can (big, collective) WE claim to know? How does (big, collective) WE access it? In recent weeks, I have been sifting through old issues of Journal of Advanced Composition, like its inaugural issue, 1.1, from Spring 1980, forty-five years ago. The table of contents was printed on the gray-blue cover:

  • Transferable and Local Writing Skills, W. Ross Winterowd
  • Cassette Commentary, Alex Medlicott, Jr.
  • A Reply to Medlicott, Karen Pelz
  • Some Theoretical Speculations on the Advanced Composition Curriculum, Richard Fulkerson
  • The Subject is Writing, Richard Gebhardt
  • Writing for the Pre-Professional Within the Liberal Arts Curriculum, Roberta M. Palumbo
  • Advanced Composition: A Survey, Michael P. Hogan
  • Review of Composition and Its Teaching, Richard Fulkerson

Listed in order of appearance in the issue, all eight contents are available as PDFs from JSTOR. The list invites questions. When was one of these last cited? Which item is Least Recently Cited (LRC)? Is citation the only meaningful (or relevant) index for use? Is Composition and Its Teaching still available after all these years? Some questions are easier, some harder. If we can agree to entertain use indices beyond citation, Which item is Least Recently Taught (LRT)? Which is Least Recently Read (LRR) by a human (not some AI agent)?

We should feather these questions out and ask them of our own work, our own collections, whether print or digital, partial or whole. Which PDF from the trove is Least Recently Read? An LRU policy might be tempted (or strictly rule-bound) to evict that PDF, to condemn it for having faded to the disappearing edge of attention’s long tail. I understand the impulse to evict what is longest-unused, the shoes with a split seam or broken lace, the pair of socks whose one mate is frayed, etc. Whether for material objects, or for digital archives of academic journals, an epistemological (and, therefore, existential) consequence looms over impending, inevitable eviction. The consequence, as I think of it, reunites for knowledge the far too commonly sequestered phenomena of repair and attention. That is, to read the Least Recently Read activates memory and patches epistemology; it invigorates a temporal bandwidth with a wider, not narrower, aperture. The least recently becomes most recently. Neither alchemical nor acrobatic nor special, this least-to-most contrapuntal might just be the practice any academic discipline needs to endure.

Gorgoylean Methods

Appealing are the sense-making motives in the Berlant-Stewart exchanges, with a nod echoic to Jenny Rice’s variation—gorgoylean methods—in Awful Archives where the generative tenets follow, 1) What is going on? and 2) What accumulates as being rhetorical figuration? and 3) How does it (fail to) add up? Not anchored entirely in story nor narrative, in description, in data nor database/collection, the gorgoylean approach hearkens maybe to positional disruption: What is for me phenomenological is for you empirical is for Earl not even worthy of inquiry.

Writing Feverlets*

Curious about her critique of Derrida’s Archive Fever, I picked up a
copy of Carolyn Steedman’s Dust: The Archive and Cultural History from
Bird Library, recalling it from another patron who had checked it out (v. sorry
about that). I deal briefly with AF in Chapter Three. Steedman
makes the point that AF is less about archives than about Derrida’s
concern for the slippage of origins (a theme in his other work) and the
inseparability of psychoanalysis from Freud (and also Judaism). She
writes, "The Foreword [to AF] carried the main argument, about Freud’s
Jewish-ness, and the contribution of Jewish thought to the idea of the archive,
via psycho-analysis" (7). Basically, Steedman is suspicious of Derrida’s
characterization of the fever (as a frenzied pursuit of origins which do not
properly exist). She complains that the concept of the fever is degraded in
translation from Mal d’Archive, and then she enthusiastically claims the
sickness Derrida mocks: "Archive fever, indeed? I can tell you all about
Archive Fever!" (17). Dust undertakes this "all about-ness" at fever’s pitch;
Steedman, all the while, works to correct (or tune, at the very least) Derrida’s
glancing consideration of the archive left behind in his treatment of other
concerns (psychoanalysis, Freud, and so on).

Continue reading →

Double-Dutch

Derrida, in Archive Fever: “For the time being, I will pull from this web a single interpretive thread, the one that concerns the archive” (45).

I am trying to bring in just enough Derrida at the end of chapter three to capitalize on his insights about origination myths (not of psychoanalysis, for my purposes, but of composition studies), about archivization as the perpetual rearrangement of data, and about the ways transclusive texts and digitization re-distribute and also re-calibrate institutional (or disciplinary) memory. This and more in 6-8 pages.

It is as if the “single interpretive thread” drawn, like a jump-rope, from the web, is held on one end by Derrida and on the other end by Brand. In this section on “How Archives Learn,” I am beginning with the overlap of archives (entering the houses of the Archons) and architecture. The Derrida-Brand skipping is double-dutch, because a second thread–from Brand–is also suspended (another thread) in this early portion of the final section. Two jump-ropes, two jump-rope holders. In their complimentary orbits, the two ropes come close to touching, but they alternate flight paths just enough to avoid touching. And yet I feel intensely the danger of getting tangled up.

As of today, I am four pages (1200 words) into the 6-8 pages I have allowed myself for the section–a necessary cap if I am to keep the chapter under 50 pp. (jeeps, when I promised myself just 35 pp.; so much for control). What remains of the section, however, is well-planned; it will be close.

One challenge has been that there is so much more more more to develop here. For instance, do we have a disciplinarily (or even a post-disciplinarily) shared theory of archivization or memory? And how important is such a thing (not only for online archives or scholarly journals, but also for the preservation of course descriptions, syllabi, listserv exchanges, and so on)? With this, I am not asking about methodologies for dealing with archives of interest to R&C (or of history and historiography, for that matter), but rather of the life cycle of a more explicit class of disciplinary materials. Is it irresponsible (even unethical) not to have greater consensus for archivization or for the “scholar of the future” Derrida writes about? Perhaps.

Next I will return to the matter of learning by squaring with a couple of propositions from Brand. Finally, there will be something on Brand’s contrast between adaptation and “graceless turnover” and also on North’s statement from The Making of… that “Composition’s collective fund of knowledge is a very fragile entity” (3)–an excerpt I work with briefly in chapter one. Maybe some of this will have to be canned later on. There is always that possibility. The chapter is, after all, building up a discussion of tag clouds, data-mining, and folksonomy, which musn’t be abandoned in the concluding section.

An Address

Today’s

Strange Maps shows a map
of ‘the island’ in Lost, and in the
discussion, there is a question about naming, an observation that it is peculiar
that the island is un-named.  In one sense, the LAT-LON coordinates name
the island, locate it, provide it with an address (I would repeat those numbers
here but for the jinx). But the island is not named (Formosa!) in the
conventional sense of toponyms.

The map itself displays layers of plausible locations (colored dots) and
zones (rings) meant to match up with events over the first three seasons of the
program. I find the map interesting because it surfaces at the same time I am
reading and (sketchily) writing about archives, tagging and keywording, what
Derrida in Archive Fever calls the archontic dimension–consignment,
the gathering and piling on of signs.

What does the map archive? And where is the imaginary map between
commencement (sequential) and commandment (jussive)?

I don’t know.  I cannot settle this yet, and I am in no hurry. Lost
is not even airing again for a couple of months, and then, only if the writers’ strike is
resolved. Nevertheless, I am–for these few minutes–taken on a detour through
the map as a museum of Lost, of a topo-nomology embedded almost entirely in television (a
domain, like many others, about which we must continuously ask, What is lost (er,
diminished) in "legitimate hermeneutical authority" (3)?).