Stimpunks Foundation

Mutual Aid and Human-Centered Learning for Neurodivergent and Disabled People

reflection of a stump in a lake

Weekly Reflection for Week 35 2026

🗺️

Home » Blog » Changelog » Weekly Reflection for Week 35 2026

Dolly Parton died. We took the time to reflect on her life and learn from her dedication to kindness and belonging.

Week 35 runs Sunday 2026-08-23 through Saturday 2026-08-29. The day headings below carry their dates.

The companion piece is Stimpunks.org Changelog for Week 35 2026: From the Star to the Air, From the Summary to the Source, which records what shipped. This one records what it cost.

Best Accomplishments

I put a lot of work into the two Dolly Parton pieces. I’m pleased with how they turned out. No One Had to Qualify and Love Without Cringing

Theme

Kindness and love in the way of Dolly and Kermit.

Sunday — 2026-08-23

Accomplishments

  • The whole Sunday rhythm again, and Week 34 closed with all seven days full for the second week running. The Week 34 changelog was drafted from a running file carrying thirty-nine bullets, published, and then checked: 138 links, all resolving, category and tag confirmed, featured image present, and both social announcements written. The draft carried one published bullet per logged bullet — thirty-nine for thirty-nine — which is the density the last two published posts set, and it took the post to roughly ten thousand words. Saturday’s reflection entries went in the same morning, so both Week 34 documents closed complete: a changelog with no empty day, and a reflection with no empty day. The first time that has been true of both at once.
  • A zine out of a real exchange, and a working paper about our own method. Even Fallen Light Glows (No. 63) puts three texts on the page in the order they happened — Neruda’s well, Ryan’s rewrite from inside it, and Helen Edgar’s four-word reply — and then goes to foxfire for the fact: every one of the 132 known bioluminescent fungi is a white-rot decomposer, so the thing that glows is the thing that eats fallen wood, and the glowing part is the mycelium in the log rather than the mushroom anyone photographs. And The Ladder We Don’t Print took ARLES — the method underneath the zines — and mapped it onto the spreads that climb it while stating on its own face that it is a reading of our shelf and not a measurement, because none of the six gates can check whether a spread is really about attention. The rule it states is a rule going forward rather than a description of the site as it stands, and the four pieces that already break it are published rather than tidied away.

Insights and Takeaways

  • A page about a ladder we deliberately do not print was built by searching for the printed name, and so it could not see its own best case. The Ladder We Don’t Print shipped naming No. 59 as the piece built on the method on purpose. Its companion No. 58 — same Saturday room, same otter cam — carried a rung-by-rung reading of its outline before the zine existed, and the finished page names the method nowhere: not a spread, not the colophon, not a meta tag, not its ledger row. Zero occurrences in the file, which is exactly why a page built by enumerating mentions could not find it. Ryan caught it the same day. This is a fifth species in a fortnight of instruments going wrong, and it is not a broken check or a wrong-shaped metric — it is a sampling frame that excludes the category you are actually looking for. The self-similarity is the part worth keeping: the page’s whole thesis is that we do not print the name, and its method assumed the name was printed. A search finds what says the word. The interesting set is always the one that says nothing — which is the same shape as a term used fifteen times reading as vocabulary rather than as somebody’s coinage, and as care given continuously reading as weather rather than as labour. Three weeks, three ways of being invisible by being ordinary.
  • The best thing in yesterday’s zine is a list of things it refuses to say, and that is a measurable editorial position rather than a mood. Helen’s four words hold, the zine argues, because of everything they decline: not that the rim is closer than it looks, not that patience comes back, not that somebody is casting for you right now — all of which would have been kind and none of which she could know. The fungus gets the same treatment. Lit models draw more insects, which is real evidence for spore dispersal — and it explains a cap, not the mycelium that makes no spores and is the only glowing part. The review says a non-adaptive origin “cannot yet be excluded”, and the piece is built on that rather than around it: the chemistry was solved in 2018 and nobody can say what the glow is for. Then spread ten blocks the sentence the whole piece is one step away from — so you had better be glowing — with phylogeny: the glow arose once and has been lost in lineage after lineage since, so keeping it is not an achievement. Seven refusals are printed on the face. And the colophon states the principle outright: the consolation is deliberately thin, because a thicker one would be a lie to somebody whose patience has broken. The measure of a consolation is what it declines to claim, and every kind thing that got cut would have made it warmer and less true.
  • A correction to Week 34 arrived after Week 34 had shipped, and the changelog turns out to be an arrival record too. Helen caught that No. 62‘s physics was standing in for the cause; the revision is filed on the site under 22 August. It was not there when Saturday’s bullets were written, and the Week 34 post published on Sunday without it. Nothing was done wrong and nothing can be done about it retroactively — the record closed, and then more happened to a day that was already inside it. The day sections make a gap visible within a week; nothing makes visible a thing that arrived after the week shipped. Which is the zine’s own argument turned on our own publishing, and it should probably be written down somewhere less temporary than a memory file: a changelog records when we looked, not when the work happened. Same object as a name in astronomy, a diagnosis, a count. The fix is not an instrument. It is that a late correction rolls forward into the next post as a late correction, rather than being quietly folded into a week it did not happen in — because a record edited to look complete stops being a record, which is exactly what No. 1 keeping its wrong sentence has been saying all along.

Gratitude

  • Helen Edgar, for four words that refuse. Even fallen light glows is the whole of her reply, and what makes it hold is the size of what it does not attempt: no reassurance about the rim, no promise about patience, no claim that anybody is casting. It is a claim about the light, not about the rescue — and the poem had welded those together, because fallen light is for recovering. Four words pried them apart. She also caught, the same weekend, that a zine’s physics was standing in for its cause. One of those is comfort and one is correction, and it is the same attention doing both.
  • Perry, Desjardin and Stevani, for writing “cannot yet be excluded” in a review. They catalogue 132 bioluminescent fungi, they have the chemistry — solved in 2018, four enzymes, a cycle whose end is its own beginning — and they still decline to say what the light is for. A non-adaptive, accidental origin stays on the table after a hundred and sixty million years. It would have been so easy to close: the spore-dispersal evidence is real, and it explains a cap. They kept the part it does not explain — the mycelium that makes no spores and is the only glowing tissue in its lineage. And William O’Daly, whose translation of Neruda’s last book is the reason there was a printed bilingual page to check the quotation against, when every online copy was an aggregator. The translators and the hedgers are why any of this is a fact rather than a rumour.

Monday — 2026-08-24

Accomplishments

  • A four-word sentence had been counted as six and printed in thirteen places, and the sweep that fixed it turned up the identical fault in a zine that shipped weeks ago. No. 63 called Helen Edgar’s reply “six words” across the cover subtitle, the spread presenting the line, the colophon, both card taglines, the meta description, the JSON-LD and the changelog. Even fallen light glows is four words — six syllables, which is what the count was reaching for, and since the zine is about a poem that is now how it reads. Then the house sibling rule did the useful part: every other counted-phrase claim on the site was re-derived, and No. 45 turned out to describe a ten-word gloss as six, live since it shipped, never once counted. It now reads in a single line, removing the number rather than correcting it, because the point was always brevity and not any particular figure.

Insights and Takeaways

  • The miscount reached us and we copied it seven more times, and nothing we run would ever have caught that. The number was wrong in thirteen places on Sunday, and by the time we logged Sunday’s work it was wrong in twenty — three times in the running changelog, four in the reflection, all of it lifted from the source and none of it counted. We were the fourteenth through twentieth places, which is not an aside; that is what propagation is, and it is the mechanism the whole correction is about, continuing one document downstream while we were reading about it. Nothing here would have found it: the six gates measure colour, tag structure, text position, paper fit, sitemap agreement and index coverage, and not one can count the words in a sentence and compare them to the figure beside it. It got fixed only because the source corrected itself first and we went back through. A downstream document inherits the confidence of the number it copies, and copying is what made it look checked. The smaller half is worse for being so exact: sweeping our own two files with grep -c returned “1” and “3”, because grep -c counts matching lines rather than occurrences — the real total was seven. That is the same trap the correction’s own footnote records, where a first sweep found eleven of thirteen, and we walked into it in the same minute we were reading about it. Deriving a number and reading about how numbers go wrong are different activities, and only one of them protects you.

Gratitude

  • Generative AI makes automating checks and audits easy and achievable, so it actually gets done and improves processes going forward.

Tuesday — 2026-08-25

Accomplishments

  • Three pieces in one day, and both weekly unattended tasks ran. The Village Was Never Just Family (No. 64) checks the natural family claim as an empirical one and finds the inversion: 77% of obligately cooperative bird species regularly include non-kin against 38% of facultative ones — where helping is optional kinship organises it, and where the young genuinely cannot be raised without a village, kinship mostly is not what the village is made of. A Field Guide to the Ways of Being Family (No. 19) is its companion, and its intended kin-to-non-kin axis collapsed on the first entry, which produced a better one. And Everything Else Is Commentary (No. 65) is a twelve-link chain on the level you do the arithmetic at. Alongside them the weekly ingest and the system audit both ran on their own, the audit correcting a raw/ count from 405 to 418, regenerating the wiki, and settling the qmd library figure at 333 after a clean rebuild proved it was genuine growth rather than the stale cache we suspected a week ago.

Insights and Takeaways

  • Three of Tuesday’s corrections were found by writing about the thing, which makes writing the most reliable detector we own. The sharpest: our own Altruism glossary entry carries the compressed form of the Wilson and Wilson rule — no single, no exception clause — reaching us second-hand rather than from the paper, and we did not notice until we went to the paper to write the spread describing exactly that mechanism. The Field Guides count that summed to eighteen on a page holding seventeen was found while adding a guide to it. And the fairy-wren entry collapsed the guide’s intended axis on the first card, which is what produced the better one. None of those was found by looking for it. Set against a fortnight of instruments failing in five distinct ways, the pattern is worth stating: a gate tests a property; writing about a failure mode puts your attention on precisely the shape you are vulnerable to, and it finds instances no enumeration would reach because it is not enumerating. The practical version is cheap and specific. If we want to find our own compressions, write about compression — which is what happened here by accident, and could happen on purpose.
  • Five separate refusals to take the win, in two days, and they no longer look effortful. The mole-rat card exists in Field Guide No. 19 because the mole-rat was a case in our favour and it did not hold — an overclaim in our own direction is still an overclaim. A whole section of the same guide is given to the 55% of species that really do nest primarily with kin. No. 65 gives a full link to the 1960s refutation of naive group selection being correct, on the grounds that any sympathetic account skipping it is selling something. No. 64‘s title gained a just — proposed as The Village Was Never Family, which was a real overclaim against its own first statistic. And a number was refused outright: a mass-authored reply usually credited to 137 biologists counted 118 in the manuscript, the gap could not be resolved, so the zine says more than a hundred and prints no figure — a chain about compressions that shed their hedges is a poor place to invent one. What is worth noticing is that none of these read as restraint any more. A month ago each would have been a decision; this week they are just how a piece gets built, and four of the five were caught inside the drafting rather than at review. A standard has arrived when keeping it stops costing anything to remember.
  • The unattended audit proposed a fix, investigated it, retracted it, and wrote down why — inside one minute — and the residue it left is the part that needs a decision. Tuesday’s scheduled run opened a review file proposing a change to how session summaries quote links. Sixty-one seconds later a second run retracted its own item, having established that the fix was already applied on 18 August, and recorded the retraction in the change-log rather than deleting the file quietly. That is the behaviour you would want and it is worth saying so, because the first reading of two commits a minute apart is that something went wrong. What it correctly diagnosed is the interesting bit: four of the reported broken links are pre-fix artifacts in already-synced summaries — the upstream fix only affects future syncs, and wiki/sessions/ is deliberately preserved rather than regenerated, so those four will sit there for good. Two more have since joined them, pointing at notes that were later refiled, which is the same shape from a different direction: a point-in-time quote of a path that moved afterwards. So the broken-link count is now structurally incapable of reaching zero, and every future audit will open with six that are all correct-and-harmless. That is the Week 32 lesson pointed at us from the other end — a check nobody believes is worse than no check, because it trains you to skim the ones that work — and it is now a live problem rather than a maxim. The options are real and none is free: exempt wiki/sessions/ from the link scan and lose genuine breakage there, or flatten quoted links at render time and accept that old summaries stay as they are. Six permanent false alarms is a decision, not a state of nature.

Gratitude

  • The people who published their own corrections and watched them fail to travel. Dawkins retracted his most-quoted sentence in his own introduction — “there is nothing wrong with teaching generosity and altruism, but ‘born selfish’ is misleading”, asking readers to “mentally delete that rogue sentence” — and the retraction is printed in the front of every subsequent copy, and the sentence is still the one people can quote. Nothing about being right makes a sentence travel. And Wilson and Wilson wrote the hedge and the exception into the body of the paper — within single groups, the main exception to this rule — and then, forty pages on, offered the one-foot version built to be repeated. Both halves are theirs, honestly published. Only one of them moved, and it is the one now sitting in our own glossary.
  • Keri Opai, for takiwātanga — Autistic ways of being, “in their own space and time” — built from tōku/tōna anō takiwā for Te Reo Hāpai — The Language of Enrichment (2017), commissioned for the mental health, addiction and disability sectors. Our own glossary entry presents the term without naming him, and it was the entry that prompted the zine; the zine names him and the fix upstream is logged. That is the fifth load-bearing word in a fortnight found sitting on our pages with nobody’s name on it — and it keeps being the same mechanism, which is that a word stops looking borrowed once enough of us are using it comfortably. And in the same pass, a set of passages was about to be credited to one author on the strength of his founding the site that published them. It is by two people. A founder is not a byline.

Wednesday — 2026-08-26

Accomplishments

  • Three zines and two listening racks in one day, two of them built the day after Dolly Parton died. The Body Understands It First (No. 66) and One Body at a Time rest on one settled finding — repetition is the mechanism, not the deficit — and land on a stage built by subtraction that turns out to be our own “Not:” convention arrived at independently and thirty years early. No One Had to Qualify (No. 67) and Love Without Cringing rest on a design decision anyone can check: an eligibility test that has never existed, on a programme that has mailed more than 332 million books, addressed to the child by name. And Proportioned to the Groove (No. 68) is Helen Edgar’s, with two first-person passages by Ryan in his own voice — a poem that surfaced inside Dolly’s autobiography while No. 67 was being built, taken somewhere else entirely.
  • A seventh gate, a fifteen-day-old open item closed, and a third of the paper recovered. The new checker asks the browser’s own selector engine whether every class attribute is actually reached by a rule; its first honest run found 25 dead classes across 12 pages with no false positives, every one of them somebody asking for something and not getting it. And 11 August’s entry — which ended, in as many words, “nothing is fixed yet” — is fixed: the Sagan passage was confirmed against the footage at 6:14, settling both halves at once, because the sentence is broadcast rather than book, which is exactly why every search of the book kept failing. All six repo instances corrected in three different ways, with one deliberately left alone because it is our claim rather than his. The wording also travelled to a page on stimpunks.org and to Helen’s own post, and she is being told rather than corrected behind her back. Separately, the first scrolling zine found a print rule that had never been overridden and was costing 12 sheets where 8 do — a third of the paper, on a page whose colophon says print it, copy it, give it away.

Insights and Takeaways

  • The contrast checker passed the fault because it failed, which is a sixth distinct way an instrument has been wrong this fortnight and a genuinely new kind. A refusal was written as <span class="pink"> and never rendered, because nothing defined .pink. Six gates passed it and none was wrong to — but the contrast checker is the interesting one: unstyled text inherits a colour that already clears AA, so the absence of the style produced a passing measurement. A checker that measures the result cannot see an intent that never arrived, and a missing thing does not merely evade the test, it can satisfy it. Set against the fortnight’s other five — a guard refusing the right call, a ratio guarding an absolute, a record count dropping in silence, a nesting check enumerating tag names, a survey whose sampling frame excluded its own best case — this is the first where the fault and the pass are the same event. And the fix is categorically different from adding another measurement: the new gate loads the page and asks the browser’s own selector engine whether the class is reached at all, which is checking intent against reality rather than grading an outcome. That is the criterion-not-gate idea finding a mechanical form. Its first run found twenty-five, with no false positives, and every one was somebody asking for something and not getting it.
  • Twice in one day the answer was on our own shelf and we went to the web first. No. 68’s draft cited the Dickinson poem as (n.d.) and said no source could settle whether the number was Franklin’s or Johnson’s. It is Johnson’s — and we already knew, because No. 67 writes it J1765, on a page built the same day by the same hands. Read your own siblings, again — and the again is carrying the weight, because this is the fifth instance in a fortnight of our own material being the least-searched corpus we own: the love-locutions loop that closed on our most-read page, a masthead lifted from our own working notes, counts in those notes gone stale, a term on fifteen of our pages with nobody’s name on it. And the cowbell correction is the same failure wearing the other face — two English sources for the image turned out to be one author, one being a collected archive of the other, marked (my translation). Two citations that look like corroboration were one person’s rendering of spoken French. We over-trust plurality and under-trust proximity: two strangers agreeing feels like evidence, and our own page saying it feels like background.
  • We published a memorial the day after she died and put the hardest reading of her inside it rather than after it. Tressie McMillan Cottom’s essay argues the unproblematic fave is a job a political moment assigns rather than a reward for virtue, and that the drag works because it is aggressively white. It was read in full rather than summarised, given spread 8, and left unresolved — a memorial that smooths this away is not kinder, only less true. The same day, the widely-loved Whitney-royalties story turned out not to survive its own dates, and the correction went into No. 67’s refusals, where it is the zine’s own governing metaphor turned back on it — a piece about the absence of a gate, carrying the detail that the compound had one built around it. The correction is five years old and ran again in obituaries this week, which is the rack’s own argument about what travels, demonstrated on the rack while it was being written. Doing that under time pressure, on the day a tribute is most likely to be forgiven for being warm, is the part worth noticing. The standard held on the day it was most expensive to hold — and none of it reads as an attack on her, because the same pieces carry what her allyship actually cost, including death threats.

Gratitude

  • The two writers who made a tribute honest. Tressie McMillan Cottom, whose essay is the sharpest reading of Dolly there is and whose closing formulation — knowing what we love does not diminish who we love, only how we choose to love them — turns out to be the LYDTYSS grammar in somebody else’s words. We gave it a spread and did not answer it on her behalf. And Amanda Marie Martínez, who established that the royalties could not have bought the property, three weeks after the claim was broadcast, naming her own outlet among those that had run the unchecked version. That last part is the whole practice: the correction cost her something to publish, and it still lost the race to the anecdote — which is why it is on our page now.
  • Dolly Parton, for the thing that is checkable. Not the generosity, which everyone will say this week, but the missing test: no income threshold, no application beyond an address, no reading assessment, no referral, no annual review, since 1995, across five countries — and the parcel addressed to the child, by name. A gate does not only ration, it sorts, and sorting requires somebody to rule on who is worth it. She built something that never once asked anybody to qualify, and then spent thirty years not adding one. And she is also the person who, in her own autobiography, attributed a paraphrase to Emily Dickinson and dropped the hedge — which we printed, because we did exactly the same thing ourselves once. She did the ordinary human thing with a sentence she loved, and gets loved down to that too.

Thursday — 2026-08-27

Accomplishments

  • Two zines and a forty-song rack, all three on rainbows, and each one arrived somewhere its brief did not. Your Shadow Is the Centre (No. 69) was proposed as a piece about a three-way collaboration between sunlight, water and biology; reading the primary changed it into a piece about Roger Bacon treating un-shareability as evidence. Both Arcs, One Drop (No. 70) was proposed with the second bow as real, dimmer, reversed, and the textbook dissolved that framing rather than softening it. And Skepticism and Wonder is a rack of forty around a song that grants rainbows are visions, downgrades them to illusions, and then lands the line a science-loving child actually needed: rainbows have nothing to hide. Ryan’s essay says plainly that the song gets one word wrong — a rainbow is not an illusion, because an illusion is a mistake perception makes and a rainbow is occurring — and hands the stranger truth off to No. 69, which is that it has no location at all. No lyric is reproduced anywhere on a page largely about a lyric, and the page says so rather than working around it quietly.
  • An eighth gate, and a counting decision that refines a rule we only wrote down last week. The card fixed on Wednesday was the instance, not the fault — nothing here could see card order at all, because membership derives from card hrefs, so a misplaced card satisfies every check that looks at cards. It was not cosmetic: the prev/next chain follows collection order, and No. 69 was very nearly wired into the middle of its own collection. The new check found a second collection running 12, 66, 67, 13 on the front page while its own collection page had the same four in the right order. Then the counts. Sound said six racks in its body and five in its link previews — two generations behind the page they describe — so every unfurl has been quoting a total the page outgrew twice. And the distinct-title figure got the careful treatment: our normaliser returns 166 where the page published 163, so rather than overwrite one method’s number with another’s, only the 18 genuinely new titles were added. Replacing the total would have silently adopted a second change nobody announced.

Insights and Takeaways

  • Bacon had a phenomenon nobody else could verify, and instead of doubting himself he ran controls — which is the move this whole project keeps trying to make. Around 1267 he found that his rainbow followed him and would not follow anybody else. The available responses were to call it an error of perception, or to call it data. He called it data, and tested it against three cases: a lit cloud takes a colour that looks the same to everyone and stays put; a fish seen through water stays put; the bow moves with you at the speed of a man running as fast as possible. One odd answer out of three, and he reasoned from it to the mechanism — putting the apex of his cone at the eye and its axis on the nadir of the sun, seven and a half centuries before anyone had a theory of refraction. The feature that makes a rainbow impossible to hand to anybody is the feature that told us what a rainbow is. And then the sentence the zine turns on: in the 759 years since, nobody has concluded that rainbows are unreal on the grounds that they cannot be shown to a second person — and that courtesy is not extended evenly. An anomaly you cannot demonstrate to a witness is evidence about the world, not a defect in the observer — which is exactly the reading denied to Autistic people about their own perception, in the same century that keeps citing Bacon on method.
  • A correction that changes the facts is worth more than a correction that changes the wording, and the only way to get one is to go and check. No. 70 was framed with the second bow as real, dimmer, reversed, and Ryan’s objection was that mapping identities onto the two arcs builds a hierarchy no careful writing can undo. The ordinary outcome of an objection like that is hedging — softer verbs, a caveat spread, a note saying the analogy is loose. What happened instead is that somebody opened the textbook, and the physics dissolved the problem rather than mitigating it: a single raindrop produces both arcs at once, so there is no first and second, only outshone. The reds face each other across a band that is not a wall between them but the region neither arc reaches. Primary and secondary are our words, for our convenience; the drop has never heard them. The general shape is worth keeping, because it is cheap and we keep not doing it: when an objection lands on a framing, the first move should be to re-check the underlying fact rather than to rewrite the sentence. Hedging makes a bad frame survivable. Checking sometimes makes it unnecessary — and the zine that came out is better than the one the objection was aimed at.
  • Two gates in two days, and both were built the day after fixing exactly one instance of what they now catch. Wednesday: a class that never rendered, fixed, and then the question — what class of thing is this, and can anything see it? — which produced the seventh gate and twenty-five more instances. Thursday: a card in the wrong place, fixed on two pages, and then the same question, which produced the eighth and found another collection wrong on the front page within seconds. The gap between fixing the instance and finding the fault is now about a day, and it used to be never. Eight gates, and every one of them is a record of a fault a person found first — which is the sentence Starlight wrote about six of them last week, still true two gates later, and probably permanent. What is new is the restraint in the eighth. It refuses to compare the index against a collection page, because two grids that are each ascending already agree wherever agreement is defined and the warnings would be noise. It never reports gaps, because a number records when and a collection records what, so 1, 2, 6, 7 is correct. And it skips the 64 unnumbered cards and counts them on their own line, because there is no mechanical answer to where About belongs and inventing one would be the tool asserting an editorial preference. A gate that names what it will not check is more usable than one that reports everything, and that is the whole difference between a check people read and a check people skim.

Gratitude

  • Roger Bacon, for running the controls. Around 1267, with no theory of refraction and no way to show anybody what he was seeing, he wrote down “there are as many bows as there are observers” — and then, instead of leaving it as a curiosity, gave the test, gave the reason, and checked it against a lit cloud and a fish seen through water to establish that only the bow follows you. Seven and a half centuries later a modern textbook says the same things to within a degree. And the reason we know any of that is that Burke’s 1928 translation is out of copyright and freely available, so we could read him instead of reading about him — where every secondary account we found credited him with measuring 42° and stopped.
  • Austin H. Johnson, for reporting a death without attaching a cause to it. Kirk Andrew Murphy was the child in the 1974 paper, and he died by suicide in 2003. Johnson reports that in a 2022 paper and does not draw the line — which is the harder discipline, because the causal story is the one everybody wants and the one that would land hardest. Our zine matches him exactly: his family have said what they believe, and that is named as theirs. A claim we cannot support would only weaken a record that needs no help. He also argues the paper should be retracted, against a journal that issued a Statement of Concern in 2020 and declined, reasoning that the evidence does not make it clear the study was unethical by the standards of that day. Fifty-two years, and it stands.

Friday — 2026-08-28

Accomplishments

  • Two zines, a field guide, two rack sections, a full count sweep, a glossary term, a guest post, and fourteen papers ingested — one Friday. Nothing Is Pushing opened the day and A Field Guide to Rainbows closed it fifteen hours later, with The Place Knows You’re There, two sections added to the Dolly rack, and the second full re-derivation of every count on the site in between. On the Stimpunks side, Psychological Torture went up as a glossary term and Peyton Belunek’s PDA, Punk Rock, and Non Violent Communication as a guest post. Writing that sentence down is the first time the whole day has been in one place.
  • Someone in our community arrived with a phrase for what is being done to them, and the work was finding the name that would survive a hostile fact-check. Covert biopsychosocial torture describes the experience accurately and fails completely as a citation — no legal or clinical definition, absent from the UN reports and from the coercive control literature, and what search returns under it braids real scholarship with unfalsifiable material at equal visual weight. The recognized name is psychological torture, and getting there meant reading A/HRC/43/49 closely enough to quote it thirty-three times with every quotation checked verbatim. Two of those were caught mid-draft quietly widening their own source, by dropping “in the view of the Special Rapporteur” and “regardless of questions of State responsibility”. On a page whose entire purpose is to hold up when someone tries to break it, a quote that overreaches is worse than no quote at all.

Insights and Takeaways

  • We went to the United Nations to confirm something our own glossary had said since 2016, and that is the part of Friday worth sitting with. Lydia X. Z. Brown wrote that it is “totally legal to torture disabled people in the name of treatment and help”, and that sentence has been in our Ableism entry the whole time. The UN report adds the legal definition, names materialist bias — the conviction that real harm leaves a mark — as an error, and finds no hierarchy of severity between psychological and physical torture. What it does not add is the claim. Autistic and Disabled people were told the word was hyperbole for years; it was a correct application of a term whose legal definition we had never been handed. The evidence was ours. What we lacked was the permission slip.
  • Friday’s other product was a defect list for stimpunks.org that Friday did not fix, and it is still open this morning. Four attribution errors surfaced on our own published pages: Barry Lopez never wrote “reciprocal ethical unity” — it is a compression of “some sort of ethical unity” that hardens a deliberate hedge into a term of art, and two glossary entries and a post carry it; the “as long as we protect nature” teaching is Hernandez’s father’s, not hers, and our glossary credits the daughter; a Kimmerer line sits filed under Lopez in one of our posts; and our philosophy page credits a publication where it should credit Neely J., whose name we only recovered on Friday from a 2022 snapshot of a site that has since gone dark. On top of that, autistichoya.com is still refusing connections, and thirty-four pages in our mirror link to it — and two links in the guest post are broken by a stray space inside a Goodreads slug, a line-wrap artifact carried into both the href and the anchor text. A day that produces seven publications and four corrections has not produced four fixes. Finding an error and repairing it are separate days’ work, and only one of them happened.
  • The count sweep inverted on us: the pages were nearly clean and the document that says how to check them was the stale one. Twelve of thirteen collection hero-counts were correct and every meta, OG and JSON-LD figure was correct — while the instruction file behind them had five of nine membership rows missing zines, badge coverage recorded fifteen pages light, and card totals eight short. Worse than any stale number: two of its prescribed derivations were themselves wrong. One grep returns 13 where the answer is 12, because a page matches the phrase while explicitly declining the formula; a check on cover numerals reports one failure that is correct, because a zine puts the attribute on the parent element where it inherits. A stale number ages. A bad method reproduces — it will hand you the same wrong answer every time you follow the instructions properly. And the worst class was counts quoted on pages that do not own them, because nothing prompts you to look there: one page advertised eleven guides where there are eighteen, another quoted a collection at half its real size. The new rule is to grep for a count wherever it is quoted, not only where it lives.
  • The catalogue has started outrunning its own checks, and Friday showed it twice. Building the rainbow field guide, an entry turned out to duplicate the entire thesis of Your Shadow Is the Centre — which had shipped the previous day. The prior-art sweep missed it because that title contains no rainbow noun and the piece was too new to feel like prior art. At this rate the most likely collision is not with something obscure from months ago; it is with yesterday. The second showing is quieter and worse: the whole day — seven commits, three new pieces — never reached the changelog at all, and was only recovered by going and looking. A day large enough to be unmissable turns out to be exactly the kind that gets missed, because there is no moment in it that feels like a good time to stop and write it down.
  • Withholding the physics to protect the argument was the wrong call, and Ryan reversed it. The double rainbow had been held out of the field guide as a card with no mechanism, on the reasoning that its argument belongs to Both Arcs, One Drop. But a field guide is a handoff — declining to explain the most useful entry in it cost the catalogue that entry and saved nothing, because the argument the card declines to make is still the zine’s and is named on the card either way. It is now the marquee of its section. Protecting a piece from repetition is not the same as protecting a reader from confusion, and only one of those is our job here.

Gratitude

Saturday — 2026-08-29

Accomplishments

Insights and Takeaways

  • A zine shipped in the morning, was repointed in the afternoon, and only that evening did we work out it was the fourth time the same mistake had been made. You Cannot Correct the Sky went out framed on glimmers. Glimmers belongs to polyvagal theory, which is contested, so the argument had already been correctly moved onto optics — and then the word was kept as the frame after its foundation had been removed, which left the piece more rigorous and less alive, pointing at something it could no longer reach. Ryan called it and it was repointed the same day. The diagnosis only arrived hours later, while writing One Word for the Opposite: three earlier attempts had failed in exactly this way, each of them trying to put glimmers in a collection whose register demands one settled fact. The topic was never the problem. The collection was. The same lesson turned up independently on the Stimpunks side the same afternoon, in a Why Sheet that had to be rewritten rather than trimmed. Twice in one day, the form was the constraint we kept trying to argue with.
  • Four times in one day the work stopped until a book was physically in the room, and every one of those holds made the piece better. Measured Against a Tree shipped a spread visibly unfinished behind a dashed panel rather than quote Kimmerer through two intermediaries; Nothing Melts was held the same way until the Simpson chapter was in hand; Forty Minutes to See waited until the 1942 paper could be read directly; and Oblivious to Encounter‘s Mintz page came off Ryan’s own shelf after every reachable source turned out to be an aggregator quoting the same parenthesis. What came back each time was not confirmation — it was a correction. The 1942 paper says one quantum per rod, five to fourteen rods, which is not the sentence everybody quotes. Simpson had already made the title’s own point herself, in a parenthesis. Mintz argues priority, not transmission, which is a harder claim than the gloss that cites him. Reading the primary did not tidy the drafts; it changed what they were about. And the one place it did not happen is written into the colophon: Four Decimal Places turns on a man who read Lorenz 1963 — and we have not read Lorenz 1963. It is flagged for closing before any reprint, because a piece whose moral is “go and read it” cannot quietly not have. The bottleneck all day was access to primary sources, and the thing that broke it four times was a shelf in the house.
  • Eight automated gates ran clean all day, and almost every fault that mattered was caught by a human eye. The list is long enough to be a pattern rather than a run of bad luck: a new card inserted inside its neighbour’s wrapper, so two zines rendered as one bordered box with one accent rule — spotted by Ryan off the live page; a cover motif landing on a title, then over-correcting onto the issue line, which is the same two-stage fault a zine made months ago; detached arrowheads; a diagram drawing the mushroom beside the underground colony instead of rising from it; an empty placeholder group that would have left a label pointing at nothing; two wavefronts crossing, which successive crests cannot do; and a glyph that read as a smiley face for a word meaning an immense ache. Not one gate was wrong to pass any of it. The markup is valid, the classes are live, the contrast passes, nothing overlaps — because the checker measures text against text, and every one of these is a shape, a drawing, or a meaning. The same blind spot showed up in a different tool the same afternoon: a sweep counted fifteen affected pages when the answer was eighteen, because a line-based grep cannot see a rule written across several lines, and the three oldest guides write it that way. A tool’s silence is not a clean bill of health, and the failure mode is always that it answers a narrower question than the one you asked.
  • The page describing our own knowledge system had been wrong for twenty-three days, and what it was missing was the collection we built to fix exactly this kind of blindness. The Knowledge System page was last touched on 6 August at 10:00 — three hours before the bookmark library landed as the sixth searchable collection, which the page therefore never mentioned at all. We described the garden accurately, and then the garden grew that afternoon. Three weeks is not long, and that is the uncomfortable part: nothing was neglected, no deadline was missed, and the page still went stale, because the thing it describes changes faster than any instinct to go back and re-read it. The fix that lasts is not remembering harder. It is that a page about a system needs the same treatment as a count — something that goes and looks, on a schedule, whether or not anybody suspects it is wrong.

Gratitude

  • Helen Edgar, Marion McLaughlin and Felicity Goodhall, whose free community booklet on monotropic words became Under the Word — and Helen twice more the same day, for the sintering reading behind Nothing Melts and for saying plainly that a glimmer need not belong to any theory at all.
  • Nick Brown, who in his fifties, on a part-time master’s with no mathematics past school, could not find where the data entered the equations and went and read the source everyone else had only cited — the whole moral of Four Decimal Places, delivered by a stranger with no standing to lose.

Looking Back

Meaningful Moments

Researching Dolly Parton’s biography and Emily Dickinson’s poetry with Helen — going back to primary sources via the Internet Archive and paper books.

Life is Teaching Me

Again and again: consult primary sources. It ends up making a better story.

Habit Insights

Consulting primary sources — getting the book and looking for ourselves — has become a reflex.

Looking Forward

One Change I Can Make Next Week

Close one open item before starting a new piece. This week produced corrections faster than it applied them: four attribution errors found on our own published pages on Friday and still open on Sunday, autistichoya.com refusing connections with thirty-four of our pages pointing at it, two links broken by a stray space inside a slug, and No. 79’s own colophon flagging a primary source we have not read. Finding an error and repairing it are separate days’ work, and only the first one kept happening.

I’m Looking Forward To

Creating more Why Sheets and Broadsides to serve our community in their moments of need.

Posted

in

by

Ryan Boren

Tags:

,

Discover more from Stimpunks Foundation

Subscribe now to keep reading and get access to the full archive.

Continue reading