A Readings Journal: Let the #BOTIs hit the floor!
Wrestling with references, resources, reading (and listening and viewing...) bookmarks, Web and offline resources, has been a growing nightmare for the past half-decade or more. BOTI may be a large part of the solution.
I've experimented with several methods. Index cards have proved particularly useful. But they're not quite as portable as I'd like, and I've been kicking around notions for how to track achieved and aspired readings in a journal of some sort. Inspirations include the idea of a Commonplace Book and Bullet Journals, the latter of which I've used to good effect for several years now. I'm looking at ways of organising a ... research journal? Sources reference? Naming this thing itself is proving Embarrassingly Difficult....
The Secret Sauce of the #BulletJournal (also known as #BuJo) is:
- An index -- this provides a quick reference to content.
- Various template pages. These can be as simple or complicated as you like, though I prefer simple. Examples are the index itself, "Future Log", monthly (and for some, weekly) overview pages, various "spreads" (contacts, references, projects, recurring tasks, activities, long-term plans and goals, exercise, diet, etc.) and the like.
- A reasonably uniform notational format for ToDos / tasks, meetings, messages, communications, scheduled and forward dates, and the like
The last gives rise to the "bullet" name, though it's the part of the journal I'm probably least faithful to.
What's proved most useful is the resilience of the format to inconsistent use of the journal. Since you're filling out the BuJo as you use it, and aren't confronted with pre-printed (and sized) pages for a given date, if you need to use two, or twenty, pages for a single day, you can. And if you skip a day, or a week, or a month, you simply pick up again with the next day (and, probably, a month overview page). The BuJo is a "no-shame", very forgiving, format.
The structure is in the index, and the ability to cross-reference pages.
A bit of recording philosophy...
Pardon the segue, but ... I don't have my usual blogging outlets (I'm working on that), I've got an awful lot of pent-up thoughts, and ... they sort of apply here.[1]
Writing is a form of recording, as opposed to signals or transmissions, which are streams. One of the characteristics of most writing media is that once you have created a record --- pencil or ink on paper, carving on stone, etc. --- it's at least inconvenient to either un-write or rearrange text. The old physical methods for doing this, literal cut-and-paste, with scissors and paste-pots, literally objectify text --- words, phrases, paragraphs, pages --- as physical objects.
Digital formats differ in that they're ultimately fungible. While digital information can be ... stubbornly ... persistent, it is also characterised by the ability to be rearranged and rewritten with tremendous ease. Something most physical media don't afford.[2] Index cards were introduced long ago (by Carl Linnaeus) to address the need for re-organisable and re-orderable information, a role they serve admirably.
A bound codex is a useful format for publishing completed material, whose structure has already been determined, or as a general-purpose journal or ledger. It's ... less immediately useful where there is a need to be able to reference information back and forth through a given text, or where an idea or concept may be added to over time. If the recording is fixed in place, then some means of flexibly cross-referencing, assimilating, and referencing the information is needed. More on that below.
But one role at which a codex works fairly well is in giving a fixed allocation for a given amount of information, and then filling and completing the page, literally. The limitation of non-, or at least, difficulty-in- reorganisation, becomes a feature.[3]
Endless Accumulation is a Bug
The problem with many technical information capture systems, especially those oriented around the Web, is that they tend to be single-silo buckets. You just keep chucking stuff into them, with no end. Browser bookmarks, bookmarking services (Instapaper, Pocket), even Ye Olde Personal List'o'Links --- just tend to Grow Without Limit.[4]
One of the useful affordances of a daily newspaper, or weekly or monthly magazine, is that while you get a pile of articles flowing in, they are grouped by time. When the occasion comes to clear out the clutter, you simply gather up yesterday's paper, or last month's (or year's, or decade's) magazines, and dump them in the recycling bin. They're already neatly containerised into a (semi-)structured object.
(Keeping or clipping out what you want may be slightly more difficult, but is also fairly readily supported.)
Digital litter ... tends to fail this. Even such basic details as datelines are often missing, let alone tracking of when you acquired a link (or how you came across it, or what you'd associated it with). I've long lamented Readability's abysmal discovery and organisational tools (there's been next to no progress in any of these). It's also exceedingly difficult to keep track of what you've read and what had significance over time. Browser history itself is rather too inclusive, and tends to uselessness.
But time-limited accumulation can be useful
Having just been through another end-of-year (and, for at least most people, a decade), there's that challenge of recapping the whatever that just was. Which means suddenly having to pull a whole slew of "what was the best of...?" out of memory.
Wouldn't it be better to be prepared?
There's a concept from system monitoring that might be useful. For logging ongoing activities, for extended periods of time, but with lower fidelity in the past, there's a system called RRDtool, which works on a circular buffer, that is, a cyclic storage space with a finite number of storage slots, for which when the last position is used, data is overwritten in the oldest record space. This enables capturing information from an infinite (or at least indeterminate) stream into a finite space.
And you can chain multiple circular buffers together: once you've filled up an every-second-of-the-minute record, you can roll to the every-minute-of-the-hour buffer, then hour-of-the-day, then day-of-the-week, then week-of-the-month, then month-of-the-year, year-of-the-decade, decade-of-the-century, century-of-the-millennium, etc., etc. Note that what I've just described -- a set of time- and calendrical- periods, is itself a set of ring buffers. That is, one minute rolls into the next, one hour to the next. There's never an overflow of seconds, or minutes, or hours, or days (modulo the occasional leap second or day), but rather, once you fill the buffer, you restart counting at the start of the period.
And the system I've just described achieves at least a 10,000 year storage capability, in a set of nine buffers with a total of 197 storage buckets, a quantity which remains constant over the entire 10,000 year duration.:
- minute * 60
- hour * 60
- day * 24
- week * 7
- month * 4
- year * 12
- decade * 10
- century * 10
- millennium * 10
The actual number of datapoints stored would be some multiple of that, but (with this rotation model) you'd still be accomplishing 10,000 years of total storage with less 200x the base level of datapoints.
And again: you could vary your storage densities -- say, keeping 1-second resolution data for a full day, or week, and minute resolution for a month (effectively storing two-levels finer than present-duration of storage), increasing total storage used ... but again, once the storage is allocated, the total amount required never changes, regardless of the duration of recorded history.
This isn't lossless -- you're achieving history by discarding resolution, but at least you get the ability to carry something forward, without being overwhelmed by your own data.[5] There's also the question of what you're going to preserve, and there's a whole set of arguments over how to "correctly" aggregate data over time.[6]
Media Saturation
One more bit of theory: there's a basic limit to how much media people can absorb (and process) in a given period. It turns out to be surprisingly limited. The total amount of textual information you can absorb in a full day reading 24/7 is on the order of a few MB. Media consumption studies have long focused on time allocation --- hours of television, or radio, or online exposure per day. There've been studies of how many words people hear or say over a day. And at population levels, there are lists of awareness --- top stories, memes, hoaxes, "Q factor" of brands and celebrities --- which show the net level of awareness on a population basis. Because Reasons there's a very strong trailing off as even the first few elements of this list are traversed. Zipf and power functions are fundamental to information and media analysis.
As a practical matter, most people likely absorb somewhere between 10 to 100 media messages over the course of a day, if not fewer, and the net total is almost certainly toward the lower end of this list.
Pragmatically: my journal format gives me 30 lines per page. Depending on how verbosely a source is recorded, this means ~10 - 30 items per page.
BOTI: Best of Time Interval
So, thinking through my frustratingly as-yet-unnamed journal, one idea is to keep a page open on a weekly or monthly basis, keeping tabs of the Best Sources of that Period. Possibly by several categories:
- Best Reads
- Best (or at least Most Intriguing) References (items suggested by others)
- Want-to-Acquire
... with others to be provided as the Muse Strikes.
(Hrm: "As the Muse Strikes" as a title? That might be better than Libro del Conosçimiento de todos los rregno....)
Over the course of a week, note which items happen to strike my fancy. Don't rate them immediately, but at the end of the week, go through and assign a (rough) ranking (stars, plusses, etc.), not necessarily trying to number items, at least not initially. At the end of a month, tally up the BoTB (best of the best), rinse, wash, repeat, through quarters and end-of-year.
There's a difference from the strict circular-buffer model in that I won't be erasing and re-writing the existing BOTI list. But I will be starting a new page and treating it as a new and freestanding entity.
Keys are:
- You're capturing items of interest for the present period.
- You Get to Fucking Move On. Rather than simply accumulating endlessly, the current list closes, and the new one starts.[7]
- Every so often you skim the cream, and review the list(s).
Further:
- I'm not capturing everything, but significant items.
- In-depth analysis and summaries are another function of the journal, captured elsewhere.
- Tracking time-usage, themes, time-to-read, and other elements, over time, is a result I'm interested in.
- Nudging myself toward more impactful sources is another element. I'd like to identify topics, authors, publishers, etc., which tend toward higher interest and payoff.
- This is an individual-curation, not a crowdsourced / collaboratively-filtered selection. Though there's a strong external-filtering process in play (many of my own sources employ this, e.g., Hacker News, Mastodon, Diaspora, Reddit, and more traditionally-curated media).
So, for a week, I tally Best of the Week, then roll to BotM, BotQ, BotY, etc.
(I'm suspecting the journal itself will be largely filled within a year.)
And I'll let you know if I'm still at this in 10,000 years.
There will also quite likely be duplicate entries --- that is, titles or references which appear multiple times (especially on recommended / desired lists). This again is a Feature, as if something keeps showing up, I should probably at least make an assessment of it.
And finally, the "BOTI" formulation has a certain je ne sais quois and or familiar association: "Let the BOTIs hit the floor."
There are other parts of this I'm still thinking through
I've already decided that a principle role of this iteration of The Journal will be To Figure Out the Structure of The Journal. So no, I don't have a Grand Scheme in Mind, yet.
I'm really hoping to have a useful cross-referencing system. I've ... run across a few ideas (there's John Locke's rather convoluted A New Method of Making Common-Place-Books), but am exploring others. In particular, keeping tabs, probably by title, author, and theme/subject, of what I'm reading (or wanting to read, or have been recommended to read, or listen to, or view, ...) is another key function.
A general format for capturing source information, references, and my own assessments, is another area of thought. Mortimer J. Adler's How to Read a Book is a strong inspiration here, though I'm ... still reading How to Read a Book....
Recommendations are, of course, welcomed.
Notes
This violates my Prime Directive of writing: don't tell them what you're writing, or why, just tell them. But that fails the Zeroth Principle: communicate effectively and appropriately. The background matters.
I'm not sure if this is an insight or a deepity. It seems useful to note, and the specific mechanisms by which digital storage are so readily fungible might also prove interesting. In particular, computer memory seems to have elements of both recording media and transmission channels, in that a memory chip consists of paired recording and sensing elements, the latter with a dedicated communication channel to the memory bus itself. Magnetic media are more analogous to audio recording or tape, with media and readers, with latency introduced by the necessity of the two to move such that a given address is positioned for writing or reading. But don't read too much into this.
Turning bugs into features is a principle mechanism of hacking.
Old-school "personal homepage" sites with "my links" are getting ... disappointingly ... hard to find.
This gets at another notion I've been looking at -- that memory, or preservation, is the denial of time. Think of some historical building, or district, or city, or document. If you're going to see that preserved, you do so by denying the option of changing it. This can lead to conflicts between preservation and development (or adaptation), in built environments. In the data space, there's the question of what resources to apply to storage and preservation. The RRD / circular-buffer model allows an extensive retention of history on a fixed budget, though it does so by preserving only on a sampled or incidental basis.
Effectively: if you're going to discard data, you're going to discard some data. Measures of central tendency (mean, median, mode) preserve the typical measure, but discard extremes. Carrying forward a set of measures may be better. This is an argument I'm vaguely aware of though I don't know the particulars.
Of course, you can always revisit old lists, and scan for Overlooked Gems and Missed Opportunities. This is a feature.