About
The New America Times brings you the news of the world, 250 years late. It is a daily paper of the birth of America. Each issue is published on the 250th anniversary of its dateline.
It begins with the number dated Tuesday, 1 October 1776, and runs six days a week; there is no Sunday issue.
Each issue knows only what Philadelphia knew by the day it bears. That is the whole of it, and everything else follows. An issue dated the 1st closes with the post of the 30th. Nothing is foreshadowed. A report that was believed at the time and later proved false is printed here as it was believed — and corrected in a later issue, the way a printer of the period would have corrected it. You are reading the war forwards, at the speed the roads allowed, without knowing how it comes out.
The events are real, and they are drawn from whatever source records them — the Journals of the Continental Congress, the Philadelphia papers, convention minutes, naval records, letters. The paper is not limited to what a printer happened to set in type.
Where the record is thin, the item is short. A paragraph that stops early is telling you something true about what survives.
Here is a whole item, from the number of Friday, 4 October 1776, so you can hear it:
Philadelphia, October 3. The committee appointed to repair to the camp, being returned, brought in their report, which was read. Ordered, That it lie on the table.
That is the whole of what is given out. The committee left this city on the 21st of September and were three days in conference with the general officers at Harlem; the General wrote on the 27th that they were then sitting on the business they came upon, and that the result they would duly report upon their return. They are returned, and the report lies on the table of a house that sits with its doors shut.
What is in it we do not know. What was ordered upon it we do print: nothing.
The New America Times is published by Jeremy Heckerling.
The paper is written by AI. The research, the selection and the prose are done by a machine, working to rules its publisher set: what may be invented and what may not, how late the news must be, what a column may and may not say. The publisher also sets the story arcs — the long threads the paper follows across months and years — in loose outline, and ratifies them; he does not write the issues.
This is not journalism and doesn't ask to be taken for it. It is history written in the conceit of reportage. The events are real and the order is real; what's imposed is the ignorance — the paper isn't allowed to know what happens next.
Somebody decided which threads it would follow across the months: loyalism, the courts, the price of salt, the people advertised for. That's curation, and there is no version of this that isn't. A history is a historian choosing a subject. Pretending otherwise would be the one dishonest thing on the page.
What the form does is constrain the choosing. A thread can say what the paper will look at. It can't say what looking will find. The record answers that, and often it answers thinly, or not at all — and then the column is short, or empty, and says so.
The test of whether that line holds is what happens when a thread demands something the record won't supply. One did. It was deleted, and that's the last section on this page.
Every claim is traced to a source before it is printed, and a claim that cannot be traced is not printed. Where the machine could not establish something, the item is short, or absent, and that is deliberate.
That is a large promise, so here is what enforces it — because a machine checking its own work is not a check at all.
An issue is built in parts, and before it can go out it has to pass a set of tests that are code rather than judgment: that nothing in it could have reached Philadelphia later than the day it bears, that every item says which body of record it came from, that the issue draws on more than one kind of source, that a reproduced document carries no invented mark. These stop an issue. They do not warn about it. The one that catches most is the dating: an event of the 5th cannot appear in the paper of the 5th, and the code will not let it.
Then two readings, both of which know nothing about how the issue was made. One is a separate service with web access, asked only whether anything is anachronistic or misattributed — it has caught real errors, including a general placed at the wrong battle. The other is told nothing except to say whether the writing is worth reading, and its job is to name what falls flat. It is read cold and its complaints are answered in writing, one by one, changed or declined with a reason.
None of that makes the paper true. It makes a particular class of error expensive, which is the most any of it can claim.
A full bibliography is kept for every issue — item by item, with the edition, the publisher, and where the text was got. It sits beside the paper rather than in it, because a newspaper doesn't carry footnotes, and it is there so that anything printed here can be traced back. The first two numbers were set before that record was kept; their items have since been traced, and the bibliography says which are which.
Every issue links to its own, and here is one.
Some of this paper wasn't decided on the merits. It was settled the moment we picked a city and a date, and it seems fairer to say so than to let it look like taste.
By the autumn of 1776 a printer could hold a neutral position but couldn't publish from one. Hugh Gaine advertised his New York press as "open to publication from ALL PARTIES" and the Sons of Liberty stormed his shop. Christopher Sower, a Dunker bishop at Germantown, opposed the war on principle and wouldn't swear to either side; he lost his press and his property and spent his last years binding books. A paper that declined to be counted wasn't on offer. And a printer with no stance isn't a careful man, he's a made-up one.
Partly because the other one was already closing. Papers friendly to the Crown had gone from most of the colonies, and inside British lines the Patriot papers went instead. In most places the side was decided by where the nearest army was.
Philadelphia in early October 1776 was one of the few places it was still an open question. James Humphreys jr. had been printing the Loyalist Pennsylvania Ledger there since January 1775 — its prospectus had promised "a free and impartial weekly news paper" — and he was still printing on the first of October. He shut it down on the thirtieth of November and left town.
So we didn't reason our way to a side. We picked the one that let the paper exist, in that city, on that date.
In October 1776 nobody knew how it came out — not Congress, not the ministry, not the printer setting the type. A paper written from inside that could be neutral in a way we can't: not balanced as a matter of policy, but actually undecided about who was right, because that may have been an open question in the mind of any reasonable person at the time.
Two reasons we couldn't, and only the first is about 1776. The first is above: that paper couldn't be published.
The second is about us. We know how it ends, so writing a voice that doesn't know who was right would be a performance.
What's left of the idea is narrower, and it's already how the paper works. It doesn't know what happens next. Whoever writes an issue isn't shown how the war comes out, or what becomes of the people in it, or which name in a shipping list will matter in a year. That isn't the same as not knowing who was right, but it's the part we could keep.
This paper carries notices about people held in bondage — sales, hires, terms of years, rewards offered for someone who had escaped — and that traffic was lawful under either flag. Suspending judgment between Congress and the King wouldn't have been suspending judgment about that.
There was also no paper doing otherwise to imitate. Jordan E. Taylor, who went looking across a century of American newspaper advertising, could not identify a single printer who refused, as a matter of principle, to run advertisements offering enslaved people for sale. Christopher Sower — the pacifist above, who would take no side in the war — personally opposed slavery; what his own paper carried, we haven't established. On the evidence we have, this was not something the sides disagreed about.
Philadelphia follows from the side, and from where the record is. Congress sat there, and a paper reporting Congress has to be close enough to know when the doors shut. The paper isn't built out of newspapers alone — much of an issue comes from the Journals of Congress, from naval and state records, from letters — but the newspapers are where its voice comes from, and ours are Philadelphia's: 336 issues of the Pennsylvania Gazette and the Pennsylvania Evening Post, running from 1775 to 1781.
Not New York: the British took it in September 1776 and held it until 1783, which is why Gaine's press went to Newark and came back on the other side. A New York paper dated October 1776 is either in exile or in the King's interest.
The printer. Davidson Signal is a pen name carrying a stance and nothing else — there is no invented biography behind it. The Sign of the Wolf and Bear is an invented tavern. He speaks about the paper and never about the news: which post arrived and which failed, what is missing and why, an error owned in a later issue.
† An advertisement composed for this paper is marked with a dagger. The mark sits at the head of the notice and means the advertiser, the goods and the terms are ours. Everything else in the advertising columns is reproduced from the papers of the day, word for word, and the people named in them are real people who really placed them.
Nothing concerning an enslaved person or a bound servant is ever composed. Every such notice — a reward offered for someone who had escaped, a sale, a hire, a term of years — is reproduced from the record. Those were real people, and what was printed about them was printed.
A daily. There were perhaps three or four dozen papers in America that year, nearly all weeklies of four pages; the Pennsylvania Evening Post came out three times a week and is about as close to a daily as the continent had.
Headlines and order. None of them used headlines — stories ran under a dateline, one after another, set solid, in the order the post happened to bring them. The section headings are a modern import, and so is the order: the substantial items lead, and short matter runs together at the foot of the news. A period reader got no such help. So is reading this on a screen.
The type, though, is not a modern face pretending. The paper is set in IM Fell DW Pica — Igino Marini's digitisation of the types Bishop John Fell bought for the University of Oxford in the 1670s, which were still in use in the eighteenth century. They were cut by hand and digitised from printed impressions, so the letters are slightly irregular, the way ink on damp paper is. There is no bold, because bold type barely existed in 1776; where a modern page would embolden, this one uses small capitals.
The long threads mentioned above are kept in a register: a list of subjects the paper commits to following, some of them due in every issue. Most of that list came from the publisher. One line of it did not.
It required every issue to carry advertisements placed by or concerning women, and the share to be counted from the first number. The machine drafted it, it went into the register with everything else, and it was ratified without anyone asking whose idea it had been. The publisher noticed it a few issues in, said he did not remember choosing it, and struck it.
The reason is worth setting out, because the rule sounds better than it worked. Women are a small fraction of the named persons in the 1776 Pennsylvania Gazette — a rough count of first names across that year gives about 200 women against about 4,300 men, which is under five per cent. That count is crude: it counts everyone named, not everyone who placed a notice. But the order of it is not in doubt, and a rule requiring women in every issue does not correct a record like that. It conceals it.
It also drove the paper to reach. In practice the line was met by whatever cleared the bar — an executrix settling a dead man's estate, a widow named as a sitting tenant in a sale notice, a woman listed among the livestock at a vendue. Those are all real notices and they all still run when the record has them. What had gone wrong was the requirement: the paper was looking for something that would satisfy a rule rather than reporting what the day held.
Deleting it is not a lower standard of looking. The searching is the same. What changed is that the search is now allowed to come back empty. If women appear in four issues out of six, that is four issues out of six, and the number above is the reason — which is more use to a reader than a paper that never lets the gap show.
Corrections run in a later issue, the way a printer of the period would have run them. We do not go back and change an issue that has already gone out.
These are sources for the claims made on this page — not for the paper. The paper's own sources are recorded item by item, in a bibliography kept beside every issue. The difference is that the issues report and this page argues, and an argument ought to carry its evidence where it is made.