A brain link carries more information per second than speech

The number the merger argument rests on. Speech moves about forty bits of information a second, in every language measured; typing, less. A brain link that beat speech would be the first time a person had a faster way out of their own head than the mouth, and the stated goal of the best-funded implant companies. No link has come close: the fastest, the 2023 speech decoders, move the equivalent of speech at half speed with errors, and they do it by reading the instructions for speech itself. Whether the brain can originate more than it already says is the open question.

What would count

A person, through a brain-to-computer link of any kind, produces information at a measured, sustained rate above forty bits a second over a session of at least ten minutes, with the rate and its method of calculation given in a peer-reviewed report, and the content not reconstructed from attempted speech alone.

Who settles it: Peer-reviewed reports stating the information rate in bits per second and how it was measured

Where it stands

Idea only. Not yet shown to work in any living animal, so there is no record to point at. Longer than 20 years, if ever.

How it could happen

First, more electrodes in more places: today's best implants read a few hundred to a thousand channels in one patch of cortex, and the companies now in trials plan thousands. Second, decoders that read several intentions at once, speech and hand and gaze together, so the channels add up rather than replace each other. Third, a way for the brain to learn a new output it never had, as monkeys learn to drive a cursor from arbitrary cells; a person would have to learn to "say" more than one thing at a time. Fourth, a measurement standard, since bits per second is only meaningful with the vocabulary, error rate and timing stated.

Where today's science would have to give way

The brain's own output rate. A 2024 analysis of every measured human behaviour, from typing to memory sports, finds that deliberate output runs at about ten bits a second however it is measured, with speech near the top of that range, and argues the limit is in how the brain forms intentions, not in the mouth or hands. If that is right, no number of electrodes helps until the brain itself can be made to intend faster, and nothing known does that. If it is wrong, the limit is engineering: tens of thousands of stable channels and decoders for intentions that do not yet have names.

What it rests on

Needs a first human of: A brain implant turns intended speech into text at conversational speed.

The last reading

Once a week every upgrade is read against the news, against what would count above. Open a line to see what that week held. Every reading of this upgrade is on this week's reading.

• 16 October 2026 Nothing new

  Nothing in the window meets the test. Paradromics claims more than 200 bits a second in a preprint and in animals, and reports a first permanent human implant in June 2026 with no human rate published; Neuralink's human cursor figures are about 8 bits a second at best. No peer-reviewed human report gives a sustained rate above forty bits a second. 'Paradromics' and 'Connexus' are watch terms worth adding.

Notes

Kept on the tree as the charter requires, with its mechanism and breaking point stated. The forty-bits figure for speech is from a 2019 study of seventeen languages; the ten-bits figure for behaviour in general is from the 2024 analysis cited in the breaking point. The 2023 speech decoders on this branch are the current best demonstrated figures and the first points of the bandwidth series, in words per minute as their papers give them; converting them to bits a second needs assumptions this record will state when the series is built.