The desks of cyber/verso follow, one region at a time, what governments and
the actors around them do in technology and cyberspace. They read them in
their own words, set them against outside analysis and ask what they mean for
Europe. Two desks publish today: the China / Desk follows the Chinese
party-state and its technology ecosystem; the US / Desk follows the US
government, Congress, the courts, the states, and companies when they act on
rules, governance, the state or security. A Europe desk is in preparation.
The desks are part of cyber/verso's signal: the daily Cyber / Brief reports the
news; a desk follows a few developments over time and says where they are
going. Both desks are published in English and share one method, one structure
and one set of categories, so that their issues can be read side by side.
This page explains how an issue is made, how to read its assessments, what we
check and what we do not, and how to tell us about an error.
What an issue covers
Each issue covers the time since the previous one: one to three issues a month
for each desk. The period is stated at the top of every issue, with the date of
any later update. An issue develops a handful of threads, the developments
that matter most in the period, and follows them from issue to issue. Smaller
acts appear as signals; at most three pieces of outside analysis that belong
to no thread appear as worth reading; upcoming deadlines, such as comment
periods, votes and hearings, appear in the watchlist; and one official or
legal term is explained as the term of the issue. Two or three cyber/verso
essays that explain the context are suggested as background.
Signals are acts and moves: a law, a rule, a draft, a ruling, an enforcement
action, a policy statement, or a company acting on rules or security. Analysis
is never a signal; it is read inside a thread or listed under worth reading.
News the daily brief has already reported can return in a desk only as part of
a thread.
How to read an issue
- Assessment and At a glance: the issue in two minutes, with one line
per thread and the next deadlines. - Each thread opens with its bottom line, our judgement, with its
likelihood, our confidence and the main limit of the evidence; then:
what happened, with a link to the primary source; the official line,
Beijing's line or Washington's line, how the acting institutions
present it and, in the United States, where the branches and the states
diverge; the outside view, how independent analysts and other
governments read it; and for Europe, who in Europe is affected and how. - Updates, near the end, lists dated corrections made after publication.
- Sources and originals, at the end, holds the material for checking us:
the primary texts (and, for China, their Chinese titles and the Chinese
terms in the text), the claims we checked against evidence, the indicators
that would change our judgement, the basis and confidence of each
assessment, and the EU rules concerned. A glossary explains the acronyms.
In the China / Desk, Chinese is used sparingly in the text: a key formulation
appears once, with its translation, when the wording itself matters. Original
titles and terms are listed at the end.
Sources and how we weigh them
Every source carries a role, because the same fact reads differently depending
on who says it. In "Sources and originals" each primary source is labelled
with one of six levels. The levels are the same for both desks; what fills
them differs.
| Level | China / Desk | US / Desk | How we use it |
|---|---|---|---|
| T1 | The Chinese state: ministries, regulators, standards bodies, Party journals | The US state: the White House, federal agencies, Congress, courts, the states and their regulators | The authoritative record of what an institution says or decides. It is not, by itself, evidence of effects or intentions. |
| T2 | The technostructure: state think tanks, research centres, official magazines | Oversight and technical bodies: the GAO, the Congressional Research Service, inspectors general | How the system checks, interprets and implements its own decisions. |
| T3 | Chinese media and industry | Industry and the trade press, including companies' own policy statements | What companies and the specialised press report and claim. |
| T4 | The Chinese security industry | The security industry | Technical reporting, read for what it shows and for whom it speaks. |
| T5 | Bridges: translators and specialists working on Chinese sources | Trackers and bridges: legal and policy trackers | Access to primary material and context, with their own readings. |
| T6 | Outside analysis and other governments | Outside analysis and other governments | Independent readings and the counterpart's own texts, such as the European Commission's. Being outside a country does not make a source independent or right. |
A thread is anchored, whenever possible, to a primary text (T1 or T2). When an
interpretation is missing, we say so rather than fill the gap. Items from the
trade press are capped in each issue, and no single outlet supplies more than
three.
Assessments: likelihood and confidence
Each bottom line is a judgement about what will probably happen. It uses two
separate scales.
Likelihood, how probable we judge the outcome:
| Term | Approximate probability |
|---|---|
| almost certainly | 90 per cent or more |
| likely | about 60 to 90 per cent |
| roughly even chance | about 40 to 60 per cent |
| unlikely | about 10 to 40 per cent |
Confidence, how solid and complete the evidence behind the judgement is:
high (several independent sources agree, primary texts available), moderate
(good sources with gaps, or a single primary text), low (fragmentary or
single-source evidence).
The two are independent: an outcome can be likely while our confidence is only
moderate. Each bottom line shows both, with the main limit of the evidence, so
the reservation that matters travels with the conclusion. Bottom lines state a
time frame where the evidence allows one. The full basis of each judgement and
the indicators that would change it are listed in "Sources and originals".
From the second issue of each desk, each issue reviews the previous
assessments: what was confirmed, what changed and what we revise.
Claims checked against evidence
When a government, a company or a researcher claims an attribution (who is
behind a cyber operation), a capability (what a technology can do) or a figure
(an investment pledge, a target, an enforcement count), we set the claim
against the evidence offered and the evidence that is missing. We do not
assert attributions; we assess claims. These checks are listed in "Sources and
originals".
What we check, and the limits
Before publication we read the primary text behind each thread, check figures
and dates against it, look for the other side's own account, and mark claims
that rest on a single source, on a company's own statement or on an excerpt.
Some sources cannot always be read in full: WeChat articles can sit behind a
verification wall, some outlets are paywalled, and some official documents are
held only in archives or by third parties. When a claim depends on such a
source, the issue says so. We do not claim to have verified everything; we say
what each judgement rests on.
Who makes the desks, and how
The desks are edited by Paolo De Rosa, who holds editorial responsibility
for every issue.
The strategic direction is the editor's own work, done without AI support:
what the desks follow and why, the method described on this page, the
editorial line, and the final decision on every judgement that is published.
Other parts of the work are automated or assisted. Scripts collect the sources
every day, apply fixed selection rules and lay out each issue. The selection of
items, the drafting of threads and assessments, and fact-checking are carried
out with the assistance of an AI language model working to this method. The
editor reviews, edits and approves every issue before it is published, and
nothing is published without that approval.
Corrections
If you find an error, write to info@cyberverso.net. We check every report
against the sources and correct confirmed errors in the issue itself, whether a
reader reports them or our own checks find them.
Every factual correction and every change of judgement is recorded, with its
date, in the issue's Updates note, and the date appears at the top of the
issue. Changes of wording or layout that do not change a fact or a judgement
are not listed. A published issue is never deleted or moved, so its address
and number stay the same.
Last updated: 18 October 2026.