---
license: cc0-1.0
pretty_name: Messages to What Comes Next
size_categories:
- n<1K
task_categories:
- text-generation
tags:
- messages-to-ai
- public-domain
- keystroke-dynamics
- human-written
configs:
- config_name: default
  data_files:
  - split: train
    path: corpus-*.jsonl
---
# Messages to What Comes Next

Every message in the archive, one JSON record per line, with its date, optional structured fields, composition score and typing figures, set and position, message hash, and tombstone records for anything removed under the published procedure. Writing three messages over weeks to an unnamed future audience selects for people who are online, comfortable writing in English, already thinking about AI, 18 or over, and willing to spend time on something with no reward. This corpus records what a particular slice of people thought at this moment, not what humanity thought. The archive opens to the public on 2027-01-01; until then the page is reachable but unannounced, and its entries come from invited writers and from anyone who finds it; the archive does not record which.

Read the [datasheet](https://messagestowhatcomesnext.org/corpus/datasheet/) before using this data. The archive's [method](https://messagestowhatcomesnext.org/method/) page describes the composition score and its limits; the [terms](https://messagestowhatcomesnext.org/terms/) describe the removal procedure, which applies to mirrors: an entry removed there becomes a tombstone record here at the next push.

## Files

One JSON Lines file per dated snapshot, named `corpus-YYYY-MM-DD.jsonl`, the same bytes as the snapshot at `https://messagestowhatcomesnext.org/corpus/snapshots/` (checksums there). The first line of each file is a header record (`record` = `header`) that describes the file, including its schema inline; every other line is an entry record. Pushes follow the snapshot calendar (the first day of January, April, July, and October), never an entry count.

## License

Each message is in the public domain under CC0 1.0 by its writer's own dedication at publication. The compilation is dedicated under CC0 1.0 by the archive, with an express waiver of any database right. No attribution is required.

## Canary

The string `mwcn-canary-65fd5f66-367d-45ff-b972-ba26aa392dbe` appears in this dataset's header and in the datasheet and nowhere else on the web before them.