Unfinished Projects Across 15 Years: Fewer than One in Ten Reached the Finish — Five Conditions That Set Closed Ones Apart

Ernest Hemingway carried one novel for fifteen years and never let himself finish it. Mark Twain rewrote "The Mysterious Stranger" for about eleven years across three versions — and finished none. The unfinished isn't always about a lack of strength. Sometimes it's about choosing the wrong form to live inside.

I. The archive in a single line

I asked an AI to read my archive and line up everything I’d ever announced as a “series,” a “book,” a “project in N parts.” I wanted to see how much of it I’d actually carried to the end. Not one episode a year, the way I’d lived it — but all of it at once, arranged by date, like a row of exhibits.

The answer was uncomfortable and relieving at once: almost nothing. In fifteen years of announced serial creative forms, exactly one closed. The rest — a title, a manifesto-preface, a numbered plan, a first chapter in full — and then silence. Once. Twice. A third time. Often enough that it stopped being chance and became a symptom.

The number first, because without it the rest is just a mood: of everything I’d ever called a “book” or a “series in N parts,” fewer than a tenth reached a final status. This text is about that symptom. And about what I understood when I saw all the dots at once, instead of one at a time across the years. Because your own pattern is hardest to see from inside: you live it one episode at a time, with years of silence between them, and each time you sincerely think it’s a new beginning, not the hundred-and-first step on the same rake you laid out yourself.

II. The pattern I couldn’t see in myself

I’m steadily strong at the opening and steadily weak at the finish — but only on one specific type of form: the serial solo fiction narrative.

The opening comes easily, almost effortlessly. I invent a title, write a manifesto-preface, number a plan, do the first chapter in full. The gesture is public, ambitious, effective. And then — silence. The second and third entries stay empty headings, like set tables in a restaurant no one came to.

The first temptation is to call it laziness or a lack of discipline. But the data says otherwise, and that’s the most important detail. Because in parallel, in those same years, I close other things with no trouble at all: operational documents, lists, manifestos, a whole knowledge base, infrastructure for a business. What I don’t close isn’t “hard.” It’s a certain type. The vessel “serial solo narrative” doesn’t hold my core. The vessel “list / manifesto / operational document” — holds it, and how.

III. What it looks like in practice

A few dots, so the pattern is visible rather than abstract.

~2010, “Personalities from Under the Blanket.” My earliest documented gesture as an author of a content project — a series of interviews with people of unusual experience. Three issues announced. One written — about hitchhiking, in full, with a signature list of ten tips. №2 and №3 — headings only. I’m 17–19, and I’m already doing exactly what I’ll do for the next fifteen years.

2013, “48 Tools of God.” A catalogue of “tools from God’s workshop,” declared in the title as forty-eight units. Actually described — one. A magnifying glass that divides a nanosecond into giga-hecto-seconds. The other forty-seven — empty numbers. The text broke off at the first not because strength ran out, but because I’d proved the concept to myself and lost interest in closing the rest. After one described tool, the brain considers the task understood, and the remaining forty-seven no longer a discovery but warehouse work.

2014, “100 Days Until Death.” A novel announced for a hundred days — a hundred texts. A full Day 1 written and headings up to the fifth. Then silence. Roughly two to five percent of what was promised. The title, incidentally, turned out honest: the project really did have few days left until its death.

2019, the corpus “for a neural net.” Here the gesture is implicit but material: I gathered a large pile of texts to one day train a neural net on it. This is 2019 — before large language models became available to anyone outside labs. The corpus assembled. Experiments on it — zero. I checked systematically: no corpus_v2, no script, not a single training run. For seven years the file just lay there, like a tin bought in case of a war it never lived to meet in the right form.

2023, “Roman’s Blank.” The file name literally says “novel.” Inside — four different-genre fragments in one wrapper: seeds for a rural tragicomedy, an operational list of notes, a marketing concept, and a prose-poem dated back to 2018. Not a novel. A writer’s notebook under a mislabelled sign.

Five dots. One gesture. Strong opening, quiet non-finish — every time on the serial solo form.

Five dots of one gesture: strong opening, empty numbers thereafter. For thirteen years I repeated the same move and each time thought it was a new beginning.

IV. What an unclosed file actually does (not what you think)

Here it’s worth stepping out of my own archive for a minute, into someone else’s dull, verified science — because it reverses the whole moral of this text. In 1927, the Berlin psychologist Bluma Zeigarnik showed that waiters remember unsettled orders better than settled ones. The unfinished seems to hang in memory as active, like a browser tab you forgot to close. For decades the conclusion drawn was: the unclosed haunts you.

And now the fresher, more uncomfortable detail. A 2025 meta-analysis in Humanities & Social Sciences Communications pulled the evidence together and found two different things under one name. The memory effect — the same “the unclosed is remembered better” — reliably does not replicate. But another, quieter effect (named after Ovsiankina) holds firmly: an unclosed action is one a person tends to resume — return to it, complete it, transplant it. In human terms: your unclosed file doesn’t hang as a reproach in memory. It pulls you to return to it — not necessarily to finish it, but to take something from it and carry it onward.

And one more number that closes the loop. Masicampo and Baumeister (2011) showed: an unclosed goal does create intrusive thoughts and steal focus — but precisely until you make a concrete plan for what to do with it. The moment you give it an explicit status, the intrusion vanishes, even if you’ve done nothing yet. So it isn’t the un-closedness that gnaws. It’s the indeterminacy of status. A file without a name is a debt. A file you’ve told what it is — a tool.

V. The exception that proves the rule

In all these years there’s exactly one case where an announced serial creative form did close — and closed twice.

2011, an anthology of Ukrainian futurism. Ninety-seven pages, a print run of over a thousand copies, a printing house in Konotop, sold by hand around the dorms. Ten years later — a reissue with an ISBN. Two editions in a decade. The only serial creative form I carried to the end.

Why that one? I counted — it breaks the five conditions all the others got stuck on. Not one. All five.

Condition	Stalled series (solo narrative)	The anthology that closed
Who's responsible	Only me. If I don't write 100 texts — my failure	Collaboration: me + co-author + twenty authors. If I don't write it myself — gathering is enough
My role	Solo author of every word	Curation: publisher, editor, seller. No text signed simply by me
Form	A long narrative that has to be grown	A manifesto-list: "WE'RE FOR / WE'RE AGAINST," a roster of authors. Lists I close
Time	"Announced and someday, gradually"	Compressed: one year, one print run, one launch. Published — contract fulfilled
Finish	A digital file you can edit forever	A physical object: a run you have to print, pick up, and sell

None of my stalled series had a single one of these five conditions. The anthology had all five. This is proof by elimination: here’s what would have had to be different for the series to close. Not “more discipline.” A different construction.

VI. Non-closing isn’t a failure. It’s an acquisition mode

When I saw all the dots together, the shame disappeared — and a different explanation appeared.

Every “unclosed project” actually worked — just not as a book, but as a seed bank. For years I’ve been pulling fragments out of these files and transplanting them into new, working forms, without even noticing the transfer.

“48 Tools” gave me the formula “to be needed and useful,” which thirteen years later became the core of how I explain value in business. An announced but unwritten issue about a journal from 2010 sprouted as a separate, already-finished line about method. The 2019 corpus “for a neural net” never trained a single model — but seven years on, the idea of “an AI that remembers me” was realised by another path, through my agent’s memory. “Roman’s Blank” gave a single prose-poem into its own finished text. The seed sprouted — just not in the bed where I planted it.

Non-closing isn’t a failure mode. It’s an acquisition mode: announce wider than you’ll close, to keep a surplus of material for future transplants. Carrier files are temporary containers. The value is in what migrates out of them, not in whether the box is neatly shut.

VII. But here a safeguard is needed, or it’s self-deception

This thesis is easy to misread and go off celebrating: “great, you can finish nothing, it’s all a seed bank.” No. And this is exactly where the text must strike itself at its weakest spot, before you do.

A seed bank makes sense only when something regularly sprouts from it and gets transplanted. If a file has given no phrase, decision, product, or new form for ten years — it’s not an incubator. It’s noise with a romantic rebranding. An archive has the right to live only when it feeds the present. The difference between an incubator and a graveyard is exactly one: whether anything comes up from it. Calling a graveyard a “seed bank” is the most refined way not to bury the dead and to carry them along as if alive — and to feed them, too.

So you need not two states (“active / abandoned”) but four. Most of the pain comes from a file sitting in the wrong column:

State	What it means	The honest action
Unfinished	Will genuinely be finished in its own form	Give it a deadline and finish conditions — or move it down
Unclosed	Has no final status; hangs as a debt	Assign a status — and the debt vanishes (Masicampo)
Seed	Container closed, core transplanted into something else	Record what sprouted, and archive the carrier
Dead	Seven years, gave nothing to anything	Bury it without ceremony. An archive isn't a hospice

The worst state isn’t “dead.” The worst is when the dead masquerades as active, because then you needn’t admit either the failure or the burial. It’s the most expensive shared flat in your head: you pay attention as rent for rooms no one lives in anymore, but the doors are labelled “in progress.”

VIII. This isn’t about creativity. It’s about me

The most uncomfortable discovery: the same gesture shows up not only in texts.

In business — the same structure: a series of projects, each ambitiously announced, partly executed, merged into the next as a seed, without the classic “sold and exited.” In civic writing — serial manifestos broken off mid-sentence. One and the same move across three different spheres of life: text, enterprise, statement. This isn’t “creative weakness.” It’s the signature of an operational mode.

And that changes the conclusion. Don’t expect the pattern to “vanish” through discipline in one sphere — discipline will just move the seed bank into another room and label the doors “in progress” again. But the pattern can be directed — by a conscious choice of form before you’ve even announced a title.

IX. Who benefits from you calling them “books”

Ask who benefits from this self-deception — because everyone has an interested party inside. For me it’s the internal PR department: a tireless editor who loves to announce a “novel,” a “series in a hundred parts,” “the project of my life,” because a big title warms you while the empty numbers under it demand nothing on Monday. A grand announcement grants the status of author today, and bills the finish to some future you, who, as we know, will pay.

So the files stay “books.” Not because you’ll someday finish them, but because the word “book” on the spine warms you more than an honest “archival container, core extracted.” As long as you call a seed bank a novel, you carry a false debt — and pay for it not in money but in a quiet background shame that fires every time someone asks: “so, finished it yet?” Renaming costs zero and writes off the whole debt in one move.

The word “book” on the spine warms you more than an honest “container, core extracted.” Renaming costs zero — and writes off the whole false debt in one move.

X. The tool: give every file a status

Let me reduce it to one mechanic — the most useful thing to take from here. Not “force yourself to finish.” Run each old “project” through one table and assign a status. The rule is one: no file may stay “active” without a next action. It’s either active, or seed bank, or archived, or killed — there’s no third option (“I’ll come back someday”), because that’s exactly what rots.

File / series	Year	What sprouted	Status	Action
"48 Tools of God"	2013	The "be useful" formula → business message	Harvested	Archive the carrier
Corpus "for a neural net"	2019	The "AI that remembers me" line	Harvested	Archive the carrier
"Roman's Blank"	2023	Prose-poem → separate text	Seed bank	Split into atoms
"100 Days Until Death"	2014	?	Dormant	Extract 3 ideas or bury
"Personalities from Under the Blanket"	2010	The list-tip format	Seed bank	Move into a working form

The power here isn’t in the table. The power is that after it, no file hangs as a nameless reproach. Each gets a column — and with the column the intrusion vanishes, even if you haven’t done the work itself yet. This is the practical translation of Masicampo: status is cheaper than completion, and it’s precisely its absence that gnaws.

XI. What I’d tell myself — and maybe you

Three things I’d have wanted to hear twenty years ago.

1. Stop trying to "finally finish the novel." If your steady strength is in lists, manifestos, and operational forms rather than solo narrative, then "finish the novel" is a recurring trap, not a goal. Take the form you close, and put the core there. The form you can finish matters more than the form that sounds prettier in a post.
2. Name the unclosed files what they are — a seed bank, not books. Then the shame disappears and a function appears. An unclosed project isn't a debt. It's an incubator you later harvest from. But only if you harvest.
3. If you want the series to close this time — break the conditions. Not alone, but with someone. Not a narrative, but a list-manifesto. Not open-ended, but compressed time. Not just a digital file, but a physical finish. That's the formula of the exception. It works — I tested it on myself once out of many.

Right now I’m writing a bigger thing about method and AI. And for the first time I’m applying this lesson consciously: I don’t ask “is there enough discipline,” I ask “how many of the exception’s conditions does this form break.” Breaks zero — so it’ll stall, as expected. Breaks three or four — so it’ll close, as expected. That’s the test. I don’t guess at my discipline. I design the container.

I didn’t know this about myself until an external witness laid my files in one timestamp row. Your own pattern is hardest to see from inside — because you live it one episode at a time, with years between them. The one who holds the whole chronicle at once is the cheapest consultant I can afford: it doesn’t flatter, doesn’t tire, and has no internal PR department invested in you going on mistaking a graveyard for a library.

I no longer ask myself: "Why didn't I finish this?" I ask something else, sharper and more useful: "What from this has already sprouted — and what is it finally time to bury?" The first question makes you a debtor before your own archive. The second — a gardener. And a gardener, unlike a debtor, knows a simple thing: not every seed becomes a tree, and part of the bed has to be dug over so the rest gets any light at all.

---