Folio — Essays, notes and marginalia

Started The Rings of Saturn on the train again, which makes four…

elin@folio.example (Elin Sandoval) — Tue, 11 Aug 2026 00:00:00 GMT

Started The Rings of Saturn on the train again, which makes four attempts. Got further this time — past the silkworms. I think the trick is not trying to work out where it is going.

The cost of a clean desk

elin@folio.example (Elin Sandoval) — Sun, 19 Jul 2026 00:00:00 GMT

I have never once cleared a desk on a day the writing was going well.

That sentence took about four years to become obvious to me, and I offer it as the whole argument in case you would rather not read the rest.

Tidying passes every test procrastination usually fails

Most procrastination announces itself. You are on a phone, or in a kitchen, or forty minutes into an argument with a stranger about a film neither of you has seen. There is a moment of noticing, and the noticing is usually enough.

Tidying does not have that moment, because every property we use to identify procrastination is absent:

It is effortful. You are moving, deciding, discarding.
It is productive. There is a visibly better state afterwards.
It is related to the work. A clear desk genuinely does help.
It is finite. It ends, and it ends soon, and then you will start.

Four for four. There is nothing to catch.

What it is actually doing

The thing I eventually noticed is that the impulse is extremely well timed. It does not arrive at nine in the morning. It arrives about ninety seconds after I open the document and discover that the paragraph I left yesterday is worse than I remembered.

That is not a coincidence and it is not about the desk. It is a state change — from I am a person who is about to write to I am a person who is failing to write — and the tidying is a way of not making it. As long as I am moving objects, I have not yet had the bad meeting with the work.

The test

The one thing that has helped is not resisting the impulse, which does not work, but timing it.

If the desk needs clearing, it needed clearing yesterday too. So: clear it, and then notice whether the urge survives the clearing. If it does — if you get to the end and find a shelf that also needs doing — that was never about the desk.

On a genuinely bad day I can now get about four minutes of this before it becomes unmistakable. That is not nothing. Four minutes is much better than the two hours it used to buy.

What I am not claiming

I am not claiming the desk does not matter. It does; I work badly in a mess and I know it.

I am claiming that the decision to deal with it is almost never about the mess, and that treating a scheduling problem as a virtue problem is why it took me four years. The fix was not more discipline. It was moving the tidying to Friday afternoon, when nothing good was going to happen anyway, and thereby removing its cover.

Went back to index cards for the first draft

elin@folio.example (Elin Sandoval) — Fri, 03 Jul 2026 00:00:00 GMT

Went back to index cards for the first draft. Something about not being able to reorder them without physically standing up makes me commit to an order much earlier, and the order is usually fine.

Walked to the harbour without headphones for the first time in months

elin@folio.example (Elin Sandoval) — Sun, 28 Jun 2026 00:00:00 GMT

Walked to the harbour without headphones for the first time in months. Twenty minutes in, the sentence I had been failing to write for a week arrived fully formed, which is irritating in a way I have decided not to examine.

On rereading

elin@folio.example (Elin Sandoval) — Sun, 14 Jun 2026 00:00:00 GMT

I have read Housekeeping four times, and I could not tell you what happens in it.

That is not quite true. I could give you the shape — a house, a lake, two sisters, a train that goes into the water and does not come out. What I could not give you is the thing I would have given you at twenty, which was a confident account of what the book was about. That account existed once. I remember having it. I could not reconstruct it now under oath.

The first pass is a kind of weather

The first time through a book you are mostly finding out whether you can stay in it. The prose is unfamiliar, the names have not settled, and a good deal of attention goes to the mechanical work of holding a cast in your head. What you come away with is less a reading than a climate — a sense of whether the thing was cold or warm, generous or withholding, whether you were welcome.

This is why first impressions of books are so unreliable and so strongly held. You have a real experience and almost no evidence. The confidence comes from the vividness of the weather, not from anything you could point at on the page.

What changes on the second pass

The second time, the plot stops being a task. You know the train goes in the lake. That frees up an enormous amount of the attention that was previously spent on bookkeeping, and it gets spent instead on sentences.

I do not think this makes you a better reader of the book. I think it makes you a reader of a different book. The first one — the one where you did not know about the train — is genuinely gone, and it was not a worse book. It was a book with suspense in it. You cannot get that back, and pretending the informed reading is simply the correct one is a way of not noticing the loss.

The gap is the measurement

Here is the part I keep coming back to.

If you reread something at long intervals — five years, ten — the book holds still and you do not. Everything that has changed shows up as a difference in what you notice. The passage that seemed like scene-setting turns out to be the argument. The character you found tiresome turns out to have been the only honest one. Nothing on the page moved.

So the difference is a measurement, and the instrument is the text. This is the only reason I keep a reading log, and the reason it is almost useless as a list of books — what I actually want from it is the annotation, five years old, that I now disagree with. That disagreement is data about me that I could not have collected any other way, because I would not have known to look.

Against the completist reflex

The counter-argument is obvious: there are more books than time, and rereading spends a finite resource on something already consumed.

I think that argument smuggles in an assumption about what reading is for. If the point is coverage — to have read the things — then yes, rereading is waste. But nobody who says this actually behaves as though coverage were the point. They do not read summaries. They do not accept a friend’s account as a substitute. They want the experience, which means they already believe the experience is the thing and not the information.

And if the experience is the thing, then a book that produces a different experience on the fourth pass is not a book you have finished. It is a book that keeps working.

I still could not tell you what happens in Housekeeping. I could tell you, fairly precisely, what has happened to me.

The quiet hour has moved

elin@folio.example (Elin Sandoval) — Tue, 19 May 2026 00:00:00 GMT

The quiet hour has moved. It used to be four o'clock; for the last month it has been closer to seven in the morning. No idea what changed. Rearranging around it rather than arguing with it.

The shape of an afternoon

elin@folio.example (Elin Sandoval) — Sat, 02 May 2026 00:00:00 GMT

I used to think of a working day as a container. Four hours in the afternoon was four hours of capacity, and if I got two hours of writing out of it I had wasted half.

This is a bad model, and it took me an embarrassingly long time to notice that it did not describe anything I had ever experienced.

Attention has a grain

An afternoon is not flat. It has a shape — usually a slope, sometimes a step. The first forty minutes after lunch are close to worthless for anything requiring judgement, and have always been. Around four o’clock something opens up and stays open for about ninety minutes. By six I can still edit but I cannot decide.

None of this is willpower. I have tested it against caffeine, against sleep, against starting earlier, against wanting it very much. The slope is remarkably indifferent to how I feel about the slope.

What I can change is what I put where. Editing at two o’clock is fine — it is mechanical, it has a clear finish condition, and being slightly dull is not a disqualification. Deciding what a piece is for at two o’clock produces a decision I reverse at four.

The cost of working against it

Most of what I used to experience as difficulty turns out to have been friction from putting the wrong task in the wrong part of the slope.

That is a much more cheerful diagnosis than the one I had, which was that I was insufficiently disciplined. It is also more actionable, because a schedule is something you can change and a character is not — or not on a Tuesday.

The trap is that working against the grain is not impossible. It is just expensive. You can decide at two o’clock. You will spend twice as long and produce something you distrust, and the distrust is usually correct. The task gets done, so nothing forces you to notice the exchange rate.

Two hours is the honest number

The thing I resisted longest was accepting how little of a day is actually load-bearing.

For me it is about two hours. Some days three. It has been two hours for as long as I have kept any record, across three cities and two very different jobs, and I have never once managed to move it by trying harder.

Accepting that number did two things. It stopped me treating the other six hours as failures, which had been quietly poisoning them — a lot of what made the late afternoon useless was the accumulated sense of having already lost the day. And it made me ferocious about what goes in the two.

Almost nothing has improved my work as much as being unpleasant about protecting a window I used to think was a failure of ambition.

What this is not

It is not an argument for working less. The two hours sit inside a longer day that is doing real, unglamorous work — reading, correspondence, the slow business of finding out what you think. Cutting the day to two hours would not preserve the two hours; it would starve them.

It is an argument against the container model. The day is not a volume to be filled. It is a landscape with a good hour in it somewhere, and most of the skill is knowing where.

Notes on the typography of this site

elin@folio.example (Elin Sandoval) — Wed, 22 Apr 2026 00:00:00 GMT

This piece is also the template’s kitchen sink: it exercises every prose component the theme ships, so if something is broken it is broken here first.

The body is a serif, and that is the whole design

A serif body face is unusual on the web and almost unheard of in a portfolio template. It is the single decision everything else follows from.

A sans-serif says interface. A serif says document. This is a document.

the argument, compressed

The practical consequence is that the sans has to be demoted somewhere, and the place it goes is the chrome — navigation, dates, tags, buttons. That split is strict:

The measure is 42rem and it is not negotiable

At the 18px prose size that lands around 74 characters per line, inside the 60–80 band that legibility research and four centuries of typesetting practice both arrive at.

The tokens live in one place, and the shell width is derived from them rather than chosen:

export const layout = {
  sidebar: '19rem',
  reading: '42rem',
  contentPad: '3rem',
  // 19 + 42 + 3 + 3 — recompute this if you change any of the three above.
  shellMax: '67rem',
};

Numerals do two different jobs

Old-style figures in prose, lining figures in the chrome. This is set once on body and overridden on the meta line:

body {
  font-variant-numeric: oldstyle-nums;
}

.meta-line,
.archive-date {
  font-variant-numeric: lining-nums tabular-nums;
}

Old-style numerals have ascenders and descenders and sit in a line of lowercase without interrupting it — which is what you want when a year appears mid-sentence, like 1971. Lining numerals are uniform cap height, which is what you want in a column where the dates have to align.

The tabular-nums half is doing real structural work in the archive: fixed-width digits are the only reason every row’s title column starts at the same horizontal position.

Inline things

Inline code like --layout-reading is set in the mono face at 0.9em — a relative size, so it tracks whatever it is sitting in rather than fixing itself to the prose size and looking wrong inside a callout.

A link inside prose is the one place the accent appears as text you are meant to click, which is why prose links are green and every other link on the site is not. A page where the date, the tag and the title were all accented would be a page of green.

A blockquote is distinct from a pull-quote, and this template treats them as different objects. This one stays inside the measure and takes a left rule; a pull-quote breaks out of the column and is set in italic display size.

Lists get a faint marker so the numbers do not compete with the text:

Old-style numerals in prose.
Lining, tabular numerals in the chrome.
Never both in one column.

That is everything. If you are reading this in a fork and something above renders wrong, the component is in src/components/prose/ and the styling is in the @layer components block of global.css.

Counted the notebooks

elin@folio.example (Elin Sandoval) — Sat, 14 Feb 2026 00:00:00 GMT

Counted the notebooks. Nineteen, of which four are full, two are half, and thirteen have between one and six pages used. I do not think this is a stationery problem.

Against the reading list

elin@folio.example (Elin Sandoval) — Tue, 27 Jan 2026 00:00:00 GMT

I kept a reading list for eleven years. At its peak it held four hundred and six titles. I have read, by my count, thirty-one of them — and of those thirty-one, twenty-six were books I would have read anyway, because something in the week pointed at them.

The list was doing almost nothing. It took me a long time to work out what the almost was.

The two people problem

A reading list is written by someone in a particular mood: usually late, usually after something good, usually feeling that they should be more serious than they are. That person adds The Man Without Qualities.

The person who has to read it is someone else. They are tired, on a Tuesday, in February. They did not agree to this, they were not consulted, and they are being supervised by a version of themselves who is not around to help.

Every entry is a small unpaid debt between the two, and the accumulated weight of four hundred of them is not motivating. It is the opposite. Opening the list is an encounter with everything you have failed to become.

What actually decides

When I look honestly at how a book gets read, the list is essentially never the mechanism. What decides is one of about four things:

It is physically here and the last one just ended.
Somebody I trust said something specific about it — not “it is great”, something specific.
It answers a question I am currently annoyed by.
I have already started it and stopping would be worse.

Three of those four are about the present. The list is a document about the past, and it competes badly.

The one thing it was for

Here is the almost.

Going back through eleven years of that list, the useful artefact was never the titles. It was the dates. I could see that in one particular autumn I added nine books about grief, and I did not remember choosing that, and I certainly did not notice at the time.

That is worth having. It is just not a reading list — it is a diary that happens to be written in book titles, and treating it as an instruction set was the error.

What I do now

A shelf, physically, of things I could start tonight. It holds about eleven books and it is the only queue. When something goes on it, something comes off, and that constraint does the work the list never did: it makes me choose rather than accumulate.

Everything else goes in a plain text file with a date and one line about why. I do not read it looking for what to read next. I read it about once a year, to find out what I was like.

Found a secondhand copy of a book I already own, annotated by someone…

elin@folio.example (Elin Sandoval) — Sat, 08 Nov 2025 00:00:00 GMT

Found a secondhand copy of a book I already own, annotated by someone else. Reading their marginalia is uncomfortably like reading mail addressed to somebody I have never met but apparently agree with about everything except the ending.

What a sentence owes you

elin@folio.example (Elin Sandoval) — Mon, 06 Oct 2025 00:00:00 GMT

There is a kind of sentence that stops you, and you go back, and on the third pass it opens — and you find that it needed every one of its clauses. And there is a kind of sentence that stops you, and you go back, and on the third pass you find that it is simply badly built.

Both are difficult. They are not the same thing.

Difficulty that belongs to the subject

Some things are genuinely hard to say. A thought with four conditions on it cannot be delivered in a sentence with one clause; compressing it would not simplify the idea, it would misrepresent it.

When a sentence is difficult because its content is, the reader’s effort is rewarded — not with pleasure necessarily, but with the actual thought, intact. That transaction is fair. The reader paid attention and received something that could not have been bought more cheaply.

Difficulty that belongs to the writer

The other kind is a sentence that is hard because its author had not finished thinking.

The clauses are in the order they occurred rather than the order they are needed. The subject arrives late for no reason. A word is doing two jobs and the reader has to hold both until the paragraph resolves, and then it does not resolve.

This is also difficult. But the effort buys nothing, because the compression was not in the idea — it was in the writer’s unwillingness to spend another hour. The cost was moved rather than removed, and it was moved onto someone who did not agree to pay it and, crucially, cannot tell from the inside which kind of sentence they are in.

That last part is what makes it a question of obligation rather than of style. The reader cannot audit you. They will assume the difficulty is theirs.

The unfair advantage of difficulty

Hard prose is also flattering, which is the uncomfortable part.

A reader who works through a difficult passage and arrives somewhere tends to credit the passage. The effort itself produces a feeling of significance, and that feeling attaches to whatever caused the effort — whether or not anything was actually there. Every writer who has ever been vague has benefited from this at some point, and most of us know it.

The test I use

Not “is this hard to read”, which proves nothing, but: can I say what the difficulty is buying?

If the answer is “it is holding four conditions in play, and dropping any of them would be a lie”, the sentence stays. If the answer is “it sounds more serious this way”, it goes — and it goes even when, especially when, I like it.

I fail this test regularly. But having it means the failures are decisions rather than accidents, and I would rather have written a plain sentence I chose than a difficult one I did not.

Tried reading at deliberately half speed for a week

elin@folio.example (Elin Sandoval) — Tue, 02 Sep 2025 00:00:00 GMT

Tried reading at deliberately half speed for a week. The comprehension gain was smaller than I expected; the retention gain, three months on, is much larger. Filing that away.

The long way round

elin@folio.example (Elin Sandoval) — Tue, 11 Mar 2025 00:00:00 GMT

There are two ways from my flat to the harbour. One takes eleven minutes and follows the main road. One takes nineteen and goes through a cemetery, past a primary school, and down a lane that smells — in November, overwhelmingly — of jasmine.

I have taken the long one almost every day for two years. I have never once regretted it. And I still feel, every single time, at the corner where the decision happens, a small pull towards the short one.

That pull is worth looking at, because it is not about time. I do not need the eight minutes. I have, on many occasions, saved the eight minutes and then spent them standing in a kitchen.

Efficiency as a default rather than a decision

What I think is happening is that efficiency has become the option which does not require justification.

Taking the short way needs no reason. Taking the long way needs one — and the reason has to be good enough to say out loud, at least to yourself, which means it has to be phrased in terms the short way would accept. It clears my head. I think better afterwards. It counts as exercise.

All of those are true, and all of them concede the point. They justify the long way by its output, which is exactly the frame that makes the short way obviously correct in the first place.

The jasmine is not an input

The honest reason I take the long way is that the lane smells of jasmine in November, and that is not doing anything.

It does not clear my head; some days it fills it. It is not an investment in a later hour’s productivity, and on the days when it happens to be, that is a side effect I did not go for. The walk is not a means. It is one of the small number of things in a day that is straightforwardly the point.

Once I stopped defending it in output terms, the pull at the corner got noticeably weaker. Not gone — the habit of justification is older than the walk — but weaker.

Where this generalises, and where it does not

I want to be careful, because this is the kind of observation that inflates.

It does not mean the long way is always better. Most efficiency is genuinely good, and I would not like to fill in a form more slowly on principle. The overwhelming majority of tasks are means, and treating them as ends is its own well-documented pathology.

What it means is narrower: there is a category of thing that is not a means, the category is small, and it is under continuous quiet pressure from a habit of thought that has no vocabulary for it. The pressure does not arrive as an argument. It arrives as the absence of one — as the sense that the short way needs no defending and the long way does.

Noticing which way round that is has been most of the work.