Every two months, one question.
Given the room a single essay never has. Free to read, sourced throughout, and nothing in it is a reprint.
Evolution Review Issue One · October–November 2026
The distance between proven and permitted
The distance between proven and permitted
When something that works does not reach the people it was built for, how many people had to say yes, and who is answerable for the time they took?
33 pages. A cover story and ten articles across six sections.
Twelve cases. Who holds the authority to say yes?
Ten graded against the issue’s own argument, including the ones it fails to explain. Two shown for comparison. Open any case.
2 Explains
OPM-MEG brain scanner
- Where it stood, September 2026
- Research use since 2021; not used clinically; approval hoped for 2027
- What holds it
- Device approval, then evidence assessment, commissioning, trust budget
- Who carries the cost
- Patients who could have been scanned clinically
- Confidence
- High
- In the issue
- Cover story
- Sources
- Wellcome, 7 Jul 2026; UCL News, Oct 2021
Self-healing concrete
- Where it stood, September 2026
- Full-scale tests from 2011; commercially available; rarely specified
- What holds it
- Tenders scored on build price
- Who carries the cost
- Whoever maintains the asset later
- Confidence
- Low
- In the issue
- Section 2
- Sources
- Ingenia, Mar 2011; NWO, Nov 2021; Cabinet Office Construction Playbook v1.1, Sep 2022
2 Partly explains
AI reading of brain MRI (Prima)
- Where it stood, September 2026
- Validated on 30,000+ studies, published Feb 2026; not in routine care
- What holds it
- Four approvals ahead of it
- Who carries the cost
- Patients far from early adopters
- Confidence
- Low
- In the issue
- Section 1
- Sources
- Michigan Medicine, Feb 2026; NHS England, Jul 2025
AI-discovered lung drug (rentosertib)
- Where it stood, September 2026
- Target to candidate ~18 months; phase III from Jul 2026
- What holds it
- Trial and approval, deliberately slow
- Who carries the cost
- People with a named disease
- Confidence
- Moderate
- In the issue
- Section 1
- Sources
- Ren et al., Nature Biotechnology, Mar 2024; Insilico, Jun 2025 and 7 Jul 2026
3 Runs the other way
Public consultation
- Where it stood, September 2026
- Routine
- What holds it
- A stage no answer moves
- Who carries the cost
- People with direct experience
- Confidence
- Low
- In the issue
- Section 3
- Sources
- Author’s argument
Post-quantum cryptography migration
- Where it stood, September 2026
- NIST standards Aug 2024; NCSC milestones 2028/2031/2035
- What holds it
- An internal decision, deferred
- Who carries the cost
- Owners of long-lived records
- Confidence
- Moderate
- In the issue
- Section 4
- Sources
- NIST, Aug 2024; NCSC, 20 Mar 2025
Data profiles
- Where it stood, September 2026
- Widespread; rectification limited to inaccuracy
- What holds it
- Nobody’s agreement needed
- Who carries the cost
- The person described
- Confidence
- Moderate
- In the issue
- Section 5
- Sources
- ICO, Right to rectification guidance
3 Does not explain
Small modular reactors
- Where it stood, September 2026
- First land-based commercial unit (Linglong One) planned for May 2025; not confirmed operating at 2 Oct 2026
- What holds it
- Engineering and unit economics
- Who carries the cost
- Whoever funds the subsidy
- Confidence
- Moderate
- In the issue
- Section 2
- Sources
- NucNet SMR database, accessed Sep 2026; World Nuclear News, 8 Jan 2026
Precision farming
- Where it stood, September 2026
- Dealer adoption forecasts persistently above actual, 2000-2025
- What holds it
- Designed for the wrong farm
- Who carries the cost
- Farmers it was not built for
- Confidence
- Moderate
- In the issue
- Section 2
- Sources
- Malone, Fiechter et al., Agribusiness, 2026
Cyber protection for the smallest organisations
- Where it stood, September 2026
- Deployed at scale; distributed by ability to pay
- What holds it
- Price
- Who carries the cost
- Sole traders and the self-employed
- Confidence
- High
- In the issue
- Section 4
- Sources
- DSIT Cyber security breaches survey 2025/2026, 30 Apr 2026
2 Compared, not graded
Optical atomic clock, naval trial
- Where it stood, September 2026
- Trialled at sea aboard the Royal Navy’s uncrewed test submarine XCal; announced Oct 2025
- What holds it
- One authority; nothing further for a trial
- Who carries the cost
- Nobody; it moved
- In the issue
- Cover story
- Sources
- Infleqtion, Oct 2025
Accessibility standards (kerbs, subtitles)
- Where it stood, September 2026
- Established; wider benefit documented
- What holds it
- Permission was imposed
- Who carries the cost
- Nobody; this one arrived
- In the issue
- Section 6
- Sources
- Blackwell, SSIR, Winter 2017; Ofcom, 20 May 2013
Three forecasts. Each scored in the first issue after its date, right or wrong.
Linglong One enters commercial operation
Likely, about 65%
By 31 December 2026. Confidence: Low.
An OPM-MEG brain scanner receives UK medical device approval for clinical use
About even, about 45%
By 31 December 2027. Confidence: Low.
Rentosertib is approved for idiopathic pulmonary fibrosis by any national medicines regulator
Unlikely, about 30%
By 31 December 2028. Confidence: Moderate.
What every issue carries
- Cases where the argument failsEach issue includes cases its own argument does not explain, and grades them in the open. Issue One has three.
- A source for every factual claimNumbered notes in each article’s fact box point to the primary source. Judgements are labelled as judgements.
- A corrections pageIt appears in every issue, including when it is empty. Issue One carries a correction to an essay published in September.
- Dated callsForecasts with a stated likelihood and a date. Each is scored in the first issue after its date passes, right or wrong.
- A three-line summary on every articleThe case, the claim and what it costs, so the whole issue can be skimmed in a few minutes.
- The data behind the figuresThe ledger of cases and the calls are published as CSV beside each issue, so the arithmetic can be checked.
The method page in each issue says how the cases were chosen, what the author declares, and how AI assistance was used in the drafting.
An argument that takes six pieces to build is never visible in any one of them.
The longer version
Between July and September 2026, four essays on The Next Evolution made versions of the same argument. The subjects were different and so were the sectors. The point underneath was the same, and the repetition went unrecognised until the fourth.
That is the cost of publishing several times a week. Each piece stands alone, and a reader who arrives at the fourth has no way of knowing the first three exist.
The Review is the slower form. Each issue takes one question, sets a long cover story against it, and gathers shorter articles that come at the same question from other sectors. Where an argument first appeared in shorter form on the Substack, the fact box at the foot of the article names the original.
The thing that was working before it was measured
What happens when a measurement is introduced to something that was already working, and the measurement changes the thing it was brought in to observe?
It also carries the first case tracker: every case from Issue One with its status updated, and the calls checked against events.
Follow on SubstackIf something is wrong, say so.
Corrections are published in the next issue, with the reader credited unless they ask otherwise, and the original is amended with a dated note.
Send a correctionQuote it, cite it, send it on.
Attribution is the only condition. Published every two months and free. There is no subscriber target, and no brief set by anyone other than the author.
Cite as
Catton, N. (2026). The distance between proven and permitted. The Next Evolution Review, Issue 1, October–November 2026.