Thoughtworks has published its Technology Radar since 2010 and now twice a year - volume 33 appeared in November 2025 - sorting technologies into adopt, trial, assess and hold from blips submitted by client project teams and curated by an internal group. An enterprise copies the artefact: the same four rings, a page per technology, published quarterly by a central architecture team of five. Two years on, teams do not consult it before choosing tools. What did the copy leave out, and where would copying the original be a mistake?

The situation Thoughtworks is in

A consultancy running many client engagements a year sees the same technology tried in many independent contexts. That is a sampling frame, and the radar is its output, not its method. Blips come up from the teams that shipped with the thing and a group curates them. The ring is an opinion earned by delivery, and the twice-yearly cadence exists because that is roughly how fast new evidence accumulates.

The rings carry narrow meanings. In Thoughtworks' own guidance on building your own radar, hold means do not start anything new with this technology — there is no harm in continuing to use it on existing projects. Assess is worth exploring; trial is worth pursuing where a project can absorb the risk.

What the copy left out

• The evidence mechanism. Five central architects produce rings from reading and vendor conversations. The artefact looks identical and its content is preference rather than experience, which engineers detect immediately. A radar is trusted in proportion to how visibly it was produced by people who shipped with the technology.
• The direction of flow. In the original, blips come up from teams and get curated. In the copy they come down from the centre, so the radar competes with the engineer's judgement instead of aggregating it.
• The ring semantics. An enterprise almost always reads hold as banned. Teams with the technology in production then either argue or go quiet, and the quiet ones stop reading. Publishing the original meaning — no new use, existing use is fine, migration is a separate funded decision — removes most of that resistance.
• The cadence discipline. Quarterly publication against the same narrow sampling frame means 3 of 4 editions have nothing new to say, and a document that changes without new evidence trains people to ignore it. At a week of five architects' time per edition the estate pays roughly a month of senior capacity a year for the churn.

What a working internal version looks like

Smaller and sourced. 30 to 50 blips, not 120, covering only the choices this organisation actually makes. Each blip names the teams who used it and links to what happened, so the ring is traceable. Rings get set in a session with senior delivery engineers rather than written by the centre, and it publishes twice a year with a visible diff. Adopt should be short: three or four blips per category is a real default, fifteen is a catalogue.

Where copying it would be a mistake

At small scale the sampling frame does not exist. A company of 40 engineers has one project per technology, so a ring on a blip encodes n = 1, which is one person's preference with a diagram around it. Prefer a written decision record per choice unless several teams have independently used the thing: the record carries the context, the alternatives and the date, and it costs an hour rather than a publication cycle.

It is also the wrong instrument for anything enforceable. Encryption in transit, an approved identity provider, a residency rule — none belong on a radar, because a radar advises and these do not admit advice. Put enforceable items in a machine-checked control and keep the radar for genuine choices, or it inherits the resentment that belongs to the mandate.

Common weak answers

• "Publish it more prominently." Adoption failed on credibility, not discoverability. A radar nobody believes is not improved by a link on the intranet home page.
• "Add a governance requirement to consult it." That converts an advisory artefact into a gate and loses the property that made the original useful.
• "Adopt Thoughtworks' radar directly." Its blips are chosen for a global audience, not for an estate whose real question is which of three message brokers to retire.