Essay ideas
Drawn from the 167 notes in TEMP/Notes Archive Aug 2026. Twenty-one of them
you had already tagged #blog yourself — most of those are here, plus a few
you tagged #journal that are stronger than you gave them credit for.
These are ranked by how ready they are: how much of the thinking you have
already done, how useful the result would be to someone who is not you, and
how well the piece would carry a custom design. The top five could be drafted
this month.
Each entry has a thesis because that is the part that decides whether an
essay works. A topic is not an argument. If the thesis line does not make you
want to argue back a little, the piece will not be worth reading.
1. Own the place you post from
Thesis. A profile is a room you rent by the minute from someone whose
interests are not yours. If you have a social media account, you should also
have a personal website — and the reason is not nostalgia, it is that platforms
change underneath you and take your archive's meaning with them.
Why anyone cares. Almost everyone feels this and almost nobody acts on it,
because the perceived cost is "learn to build a website". You just did it in a
day, for about 36 credits a month, and you can show the whole thing.
What you already have. Cherry (share-y) — Personal Website CMS is the
core: "All these platforms change and they don't feel like MY place." Add
Personal Software ("software that people own (again)"), Filter, and the
observation in Third Place that platforms are splitting into tribes rather
than converging on a winner.
Design. The strongest possible version is that the post is its own
evidence — the page loads in under a second, makes zero third-party requests,
and says so. A small live readout of what the page just cost the reader (bytes,
requests, origins contacted: one) would land harder than any paragraph.
Length. 900–1200 words.
2. A stopwatch, not a streak
Thesis. Streak-based habit apps are badly designed for the thing they claim
to help with. A streak punishes the relapse harder than it rewards the effort,
so the first slip deletes the evidence of everything that came before. Measure
the interval instead: how long since the last one, how long was the longest
gap. Nothing to reset, nothing to lose.
Why anyone cares. This is a real, original product-design argument with a
personal stake — you wrote it while trying to quit smoking, and the reason you
wrote it is that the existing apps failed you specifically. Anyone who has ever
broken a streak and deleted the app will recognise themselves.
What you already have. Smoking Stopwatch is almost a spec already —
laps, best lap, pausing during sleep so daytime abstinence is compared fairly.
That sleep detail is the tell that you actually thought about it.
Design. The obvious one, and it would be genuinely good: a live interval
counter in the article, plus a lap list. It argues the point better than prose
can. This is the piece most worth building an interactive island for.
Length. 700–900 words.
3. Designing with AI after eighteen years
Thesis. The interesting change is not that the tool is faster. It is that
for the first time you are working with a tool whose output you cannot fully
specify, cannot file a bug against, and cannot ask for support on — and you
still have to ship work you will sign.
Why anyone cares. Every designer is currently having this experience and
very few have eighteen years of before to compare it against. You also have the
concrete outcome: you shipped your first App Store app last month after
designing other people's for nearly two decades.
What you already have. How I work with AI as a designer, Ther's a vibe
to vibecoding ("it can get attached to a small aspect of your prompt and start
spiraling around it as a loose north star" — that is a real observation, keep
the phrasing), Using AI ("AI makes it much easier to BUILD tools that do
things reliably" — that reframe is the whole argument), A good time to try
again, and the Loop Launch Posts material.
Design. Restrained. Maybe one annotated prompt-and-result pair shown
honestly, including a failure.
Length. 1000–1400 words. This one could also be a series.
4. The worst software we use is the software we cannot opt out of
Thesis. The interface between a citizen and the state is the one piece of
software nobody may refuse, nobody may switch away from, and nobody is designing
with any care. The data already exists — the law requires it to be stored. What
is missing is not capability, it is that no one treats this as a design problem.
Why anyone cares. Everyone has a story about a government portal. Very few
people writing about it are actual interface designers, and you are one, with a
specific country to point at and Estonia as the counterexample that proves it is
a choice rather than a constraint.
What you already have. H2SI — Human to System Interface, more or less
whole. The strongest line is the comparison: this software is dogshit "compared
to the incredible software that picks our brains" — the same industry builds
both, and puts its best people on the one that harvests attention.
Careful with. The note also argues messaging and LLMs should not be
for-profit. That is a much larger claim and will swallow the essay. Either cut
it or make it the subject of a different piece.
Design. One real Romanian government form, redesigned, shown side by side.
That single image would be the most shared thing on the site.
Length. 1000–1400 words.
5. A little hope, please
Thesis. The people best positioned to build things are spending their
credibility on doom. The criticism is often correct and the conclusion is still
wrong: the open internet is still open, the protocols are still open, hardware
has never been cheaper, and the tools got dramatically better this year.
Why anyone cares. It is a genuinely contrarian position right now, argued
by someone who is not selling anything. The note is voice-transcribed and rough,
but the argument underneath is clean and you clearly mean it.
What you already have. A little hope, please?, plus If the industry is
disrupted that doesn't mean your life is too — which contains the sharpest
practical advice in the whole archive: when the industry is in crisis your
bosses are distracted, your tools just improved, and reorienting toward people
and real needs is where you still have an advantage.
Design. Deliberately plain. An optimistic essay that arrives over-designed
reads as a pitch.
Length. 800–1000 words.
6. Designing for a user who cannot read
Thesis. Building a game for your cat removes every crutch interface design
leans on. No instructions, no onboarding, no ability to ask what went wrong,
and a user whose reach is a physical fact — she cannot hit the top of the
projection, so anything up there is decoration. Constraints that severe are
clarifying.
Why anyone cares. It is funny, it is short, and it teaches something real
about designing for capability rather than intent. It is also the most
likeable thing in the archive — a good third or fourth post once the site has
a voice.
What you already have. Sisypuss is detailed and observed: existing cat
videos are made for iPads, so on a 3.5 m projection the mice are larger than the
actual cat and move where she cannot reach. That is a genuine UX finding.
Design. An interactive demo of the thing moving in the reachable band
versus the unreachable one. Playful.
Length. 600–900 words.
7. Muscle memory is a feature you do not own
Thesis. When a beloved app is redesigned, the loss is not aesthetic. It is
that years of accumulated motor knowledge are deleted without consent, and the
people who lose most are the people who used it most. Loyalty is taxed.
Why anyone cares. Universally felt, rarely stated precisely, and you can
state it precisely because you design the things that do it.
What you already have. RIP Currency — short, but it has the whole shape:
you did not mind the subscription, you minded that the muscle memory went.
Separating those two complaints is the essay.
Design. Short and sharp. Could be one of the first pieces — a good site
needs some pieces that are 400 words and land.
Length. 400–600 words.
8. I am a metalhead who lives in a walled garden
Thesis. The open-versus-closed argument is not really about openness. It is
about trust, and both models ask you to trust someone — a company, or a
community. Admitting you find comfort in the walled garden while believing in
open source is not hypocrisy, it is an accurate description of the trade.
Why anyone cares. Nearly everyone in tech holds both positions and almost
nobody says so, because each tribe punishes the admission. The metalhead
analogy is yours and it is good: "I still consider myself a metalhead, but my
Spotify top 100 wouldn't really reflect that."
What you already have. The iPhone is still fun and will be a great
interface for a long time, plus It's not personal anymore (Apple used to make
a personal computer; it now feels like a shared or cult one) and Fanboy.
Careful with. Three notes here shade into Apple commentary, which ages in
weeks. Keep the essay about trust, not about the current product line.
Length. 900–1200 words.
9. The eight-hour day is a solution, not the solution
Thesis. Working conventions are old answers to old problems that hardened
into defaults. They are not wrong; they are one option presented as the only
one. People who can afford to run the experiment owe it to everyone else to
publish what they find.
What you already have. Social Constructs & Conventions, essentially
complete, including the honest concession that the conventional setup genuinely
works for many people and jobs. That concession is what will make it credible.
Needs. Your own data. You have worked three days a week at $3000/month
while building your own things — that is the piece. Without a concrete account
of what you actually did and what it cost, it stays an opinion.
Length. 900–1200 words.
10. I bought a Leica so I would have a reason to see people
Thesis. A camera is a social prosthetic. The photographs are the excuse;
the point is that owning a good one makes you arrange the occasions.
Why anyone cares. It is the most quietly personal idea here that is still
comfortably publishable, and it is true in a way people recognise.
What you already have. MakeFuture.Gallery has the line the whole essay
hangs on, and you wrote it as a throwaway: photos of "moments in which I
interacted with people, which is not something I do often. ha ha". Also My
relationship with photography — no photographs displayed anywhere in your
childhood home except two small headshots in a room nobody used, and then your
own portrait ten times larger after you left for college. That detail is worth
the whole piece.
Design. Obviously a photo set rather than an essay — the format the blog
already supports. Would also be the first real use of the photo pipeline.
Length. 600–900 words plus photographs.
Also strong, less ready
- We need tech CEOs like the car industry — Rimac and Koenigsegg visibly
love the object; tech has operators and ideologues. Good hook, but the note
wanders into Musk and Altman and loses its shape. Needs one clear claim.
- Everyone has an opinion on design — The challenge of sharing design
work is a list, not yet an argument. The argument wants to be: whose
opinions you are obliged to take seriously, and why "entitled to an opinion"
and "entitled to be accommodated" are different things.
- Little tools, fixed price — simple tools, one-time payment, no
subscription. You shipped OPUS Loop on exactly this thesis, so wait until you
have sales data. Then it is evidence rather than preference.
- Own things and lend them — the inversion of "own nothing and be happy".
You have a real inventory (motorcycle, Leica, PS5, power tools) and a real
taxonomy of lending — indefinite, on occasion, multi-day. That taxonomy is the
interesting part.
- The most invisible part of design is development — currently one line.
It is a good line.
Handle with care — your call, not mine
Four of the best-written notes in the archive are ones I would not publish
without you deciding deliberately, twice.
- Frica de imaginar is, honestly, the most accomplished piece of writing in
the whole export. The idea — that you are afraid of the moment an imagined
good thing is about to become real, because then it stops being imaginable —
is genuinely original and pairs with I want money, because I want to live in
the future and Am nevoie. It is also explicitly therapy material.
- Your father. Vreau sa miros putin ca Tata, and the line in the tech-CEO
note. Handle on your own timeline.
- I create personal Gods — good idea, and the note file also contains three
phone numbers, one labelled neurochirurgie. Whatever else happens, do not
copy that file's contents anywhere.
- What if God talks to me through icons — involves drug use, and names
people.
None of these are off limits. They are just the ones where "it would make a
great post" is the wrong reason to decide.
Two things to settle first
Language. Roughly a quarter of the archive is in Romanian, and it is not
the throwaway quarter — Frica de imaginar, Suntem Unu and several journal
entries are in Romanian because that is where the thinking happened. The site
is currently English-only. Worth deciding whether you write in both, and if so
whether a Romanian piece gets a translation or simply stands on its own. The
schema would need a lang field and the feeds would need to declare it; small
work, better decided before there are thirty posts.
Sanitising. The notes contain phone numbers, friends' and collaborators'
full names, a Slack-style member list, financial details ($3000/month, 20K in
the bank, a 50K loan) and one hospital contact. TEMP/ is now in .gitignore
so none of it can be committed by accident — but the same care is needed when
lifting a passage into a draft.