About

Why this exists

I grew up between two groups of people who could not hand anything to each other. The older ones talked about Afghanistan constantly — a place they remembered, argued about, and were homesick for. The younger ones, the kids around me in America, did not know what any of it was. Not the history, not the places, not the names. There was no malice in it; nobody had given it to them.

It is not as simple as one generation holding something the next never received, though. Often the older ones did not have it either. Afghanistan has been at war for most of the past fifty years — the timeline on this site is not a cheerful document — and war is hard on the ordinary machinery by which a country knows itself: schools, archives, libraries, records, the unbroken habit of writing things down. What crossed the ocean was memory. Memory is vivid and partial, it argues with itself, and you cannot look anything up in it.

I am in the middle. Old enough to have heard the stories, close enough to the younger ones to see that the stories are not reaching them, and aware that the stories were never the whole of it.

An encyclopaedia is an unsentimental answer to that. It does not require anyone to sit and listen. It is there when somebody finally wants to know, and it will still be there after the people who remember are gone.

It is also a gift, and I should say who to. My four children — Eren, Talia, Issac and Hana — who are Afghan and American and have the same gap I had, with less around them to fill it. They will want to know eventually, and I would like there to be something to find.

And then the people this is actually about. The site is in Pashto and Dari, not only English, because leaving Afghans in Afghanistan out of an encyclopaedia of Afghanistan would be absurd. Those of us in the West are the ones who can already reach almost anything we want in English; the people who need this most are the ones who cannot. I only read elementary Pashto and do not read Dari at all, and the site is in Pashto and Dari anyway, for that reason.

Twenty years ago I wrote this

I asked myself once, not too long ago,
How it was they could smile and be so proud
To express out loud
This cloud of despair they've allowed
To be cast upon this place they call home.

Oh they smile when they see a child lost in this world
A rifle, a green bandana, a blood stained shirt
A malnourished child stumbling through the dirt
A future without arms, legs, or a voice to be heard
Oh but they smile with pride and proclaim that's my child

How this empire of greatness whose impact changed fate
From Buddha, the Turks, and Alexander the Great
Fell to such a state
Of despair, hopelessness, and outright hate
Quite sadly amuses me to this date

How can it be that you see such beauty in your own misery?
I am just one man.
Please help to set free
This place you call home from its disease.

Mostly it was exasperation. Give me a break. The pride was the part I could not get past — loudest among the people furthest from the place, and not obviously attached to knowing anything about it.

This is what I eventually did with that.

What I brought to it

I have a history degree from Cal State Long Beach, finished in 2005. I studied history because I liked social studies in elementary school and never stopped. Afghanistan was the one subject I could never get at: there was little in English, and a country at war for decades is not producing much scholarship.

I meant to teach the subject — I got through the first year of a two-year single-subject credential before deciding the classroom was not for me. So this is the second time I have tried to hand history to people who do not have it, and the first attempt is where the habits on this site come from: context matters, decisions compound, shortcuts have consequences. A claim is worth what its source is worth, and a reference work that will not show you where something came from is asking you to take its word. Every entry here cites.

Programming was the hobby that became the job. I started on an Apple IIe in BASIC as a kid in the nineties and kept at it, took as many computing courses as Cal State would let me take as electives — short of a minor, and I have no computer science degree — and went at it professionally once I left teaching behind. Since then it has mostly been backend work in places where being wrong is expensive — banks, loan servicing, e-commerce at scale. The parts of a system that only get noticed when they stop.

This is also not the first thing I have run for years without being paid for it: in college I took over an abandoned open-source project and spent a year and a half keeping it alive for the people using it. It is a habit. I get brought into things that are minimally defined, under time pressure, carrying technical debt, or mid-migration, and my job is to make sense of the mess and leave something others can build on.

How this is made, and what I will not pretend

The entries are written by software. A pipeline proposes a topic, researches it against sources on the open web, writes it, checks each claim back against the source it came from, translates it into Pashto and Dari, links it to the entries around it, and publishes — on a schedule, without me. A language model does the writing. The rules it works under I wrote by hand, and most of them are rules about restraint.

None of it runs on somebody else's API. The models are on a machine in my house, coordinated by a thing I wrote called cortex that hands one graphics card back and forth between them so a translation and a research pass do not collide. I would rather own the awkward version than rent the convenient one, and a project meant to outlast my attention should not have a meter running on it.

Nothing on a page that the page cannot account for. Every claim carries a citation. Every translated passage says plainly that it is machine translation from the English. Every scanned map names its author, date, licence and origin — a map that cannot say those things is not published at all.

Nothing that can be rebuilt is kept. Compiled dictionaries, fetched posters, web-sized copies of scans: none of it is committed, because all of it can be made again. What is kept is what cannot be reconstructed — the decisions.

It is static HTML and stays that way. The whole site is browsable from a folder with no server at all. A page gets JavaScript only if a particular feature earned it, and each one has to work with scripts switched off first. No account and no cookie banner.

When something is uncertain, measure it. This project keeps a folder of its own defects, and the abandoned ideas are in it too, with their numbers. Aligning the English and the Pashto word by word was tried and dropped because the measurements said it would be wrong about a third of the time. A confident wrong answer is worse than no answer.

And the thing I will not pretend: machine-written entries are confidently wrong sometimes, and I am not the person who can catch it in every language.

Pashto is my first language, but I should be precise about what that means here. I left Afghanistan young and learned it second-hand, at home, from my mother — which bounds it twice over. The vocabulary I got is a household one, and a woman's. Men and women in Afghanistan largely occupied different parts of society — politics, government and education were my father's world and not my mother's — and it was my mother I learned from. It is also pitched at the age I was when I left, so I read Pashto at roughly a child's level. I can follow a page and tell you broadly what it says. I could not tell you whether a sentence about a nineteenth-century treaty has reached for the right word.

Dari I do not read at all.

And I should say plainly that the translations are not good enough. I have rebuilt that part of the pipeline many times over — how a page is broken into sentences, which model does the work, the layers of checking that catch it mangling a name or losing a citation — and what is on the site is the best I have managed so far. A Pashto or Dari reader deserves better than this. I am sorry it is not better, but this was the best that I was able to do.

That is exactly why every claim is cited — so you can check it rather than take my word for it, or its word.

If something here is wrong, I would rather know: [email protected].

Elsewhere

GitHub, and most places, as wayubi.