Rules
Part of Platform trends: what beginners should know
Platform trends guide: history, examples and current context
Platform trends guide to a watch you can keep: an intake rule that excludes most things, sources ranked by what they carry, and a weekly half hour.
A trend watch is easy to start and almost impossible to keep, because the work is unbounded by default and the reward is invisible. Most people either read everything for two weeks and then stop, or read nothing and find out late. This page describes a watch that fits in half an hour a week, with explicit rules about what enters it, what leaves it, and when you are allowed to do nothing.
What to take away
- A watch is defined by what it excludes. Without a written exclusion rule it grows until it is abandoned.
- Sort what you read by who wrote it and why, not by how interesting it is. Provenance predicts usefulness better than subject does.
- The output of a watch is a short written record, not a feeling that you are informed. If nothing is written down, the watch did not happen.
The intake rule
Something enters the watch only if it could change a decision you have already made. That is the whole rule, and it is strict enough to exclude most of what crosses your desk.
Applying it requires knowing your own decisions, which is why the first session is not reading at all. Write down the five or six standing commitments your operation runs on: where your audience comes from, which surfaces you publish to, which tools you depend on, what format you produce, and what you have decided not to do. Anything that could disturb one of those is in scope. Anything else is entertainment, which is fine, but it is not the watch.
Sources, ranked by what they can carry
| Source type | Good for | Weak at | Read it |
|---|---|---|---|
| Operator's own documentation and policy pages | What the operator commits to in public | Timing, weighting, anything internal | On a schedule |
| Operator's engineering or research writing | Mechanism, vocabulary, direction | Currency; often describes an older system | When it appears |
| Standards and government guidance | Obligations that apply to you | Anything product specific | Twice a year |
| Trade press | Knowing that something happened | Why it happened, and whether it matters | Skim, never quote |
| Practitioner posts | Hypotheses and early symptoms | Everything else | Weekly, as a queue |
| Aggregators and newsletters | Coverage | Judgment, which has been done for you | As an index only |
Almost everything in the middle rows is gray literature: published outside formal channels, unreviewed, and often undated, which is not a reason to ignore it but is a reason to record where it came from. The order is deliberate and it is roughly the reverse of how most people spend their attention. Practitioner posts are the most engaging and the least reliable; the reasons a practitioner claim decays so fast explain why they belong in a queue of hypotheses rather than in a record of facts.
Government and standards material is the row people skip. It is dull and it changes slowly, which is exactly what makes it worth a rare visit. NIST's own introduction to getting started with its cybersecurity framework is the kind of page that is still accurate in three years, which almost nothing in the trade press will be.
The weekly half hour
Split it in three.
Ten minutes on the queue. Skim what accumulated. For each item, one of three verdicts: discard, note, or investigate. Most items are discarded, and if fewer than four in five are, your intake rule is too loose.
Ten minutes on the noted items. For each, write one sentence saying which of your standing commitments it touches and what would have to be true for it to matter. Most die here, which is the point.
Ten minutes on one investigation. Only one. If two things need investigating, the second waits a week, and if it still needs investigating then, it was real.
What leaves the watch
Items leave three ways, and naming them stops the list growing forever.
Resolved. The thing happened, you did something, it is now part of how you operate. Move it to the record.
Expired. Ninety days passed with no further signal. A change that mattered would have produced something by then. The tests that separate a durable shift from a passing one are what you apply before deciding this.
Absorbed. Two items turn out to be the same item wearing different words. This happens constantly and merging them is most of the value of writing things down at all.
The record
One file, appended to, never rewritten. Each entry carries the date you noticed, the source, the commitment it touches, and what you decided. That is four fields and it takes a minute.
The reason for the record is not the record. It is that in six months somebody will assert that a change happened, and you will be able to check whether you saw it at the time. Without that, every argument about what changed is an argument between two memories. The same discipline applied to product changes is the change register, and the two files can be one file if that suits you.
A watch also has to survive the week when nothing happens, which is most weeks. Doing nothing and writing "nothing" is a legitimate output, and it is how you avoid the trap of manufacturing significance because the slot was in the calendar. When something does happen, the question of how much of it is a cost you will actually bear is answered by looking at whether it has reached you yet, not by the announcement.
Common questions
How many sources is enough?
Fewer than you think, and the number matters less than the mix. One operator page per surface you depend on, one standards source for your obligations, and one practitioner queue is a complete watch. Adding a fourth newsletter adds coverage of the same events.
What if I miss something important?
You will, occasionally, and the watch is not insurance. Its purpose is to reduce the number of things that surprise you and to shorten the gap when one does. Anything that would be catastrophic to miss belongs in a dependency inventory with an owner, not in a reading list, and why the notice period is always shorter than it looks explains the difference.
Should the watch cover competitors?
Only as a symptom source. Watching another operation change tells you that something moved and nothing about what. You see their output and never their inputs, which is the same limit that makes outside evidence about a system so weak.
Is a half hour really enough?
For a small operation, yes, provided the intake rule is doing its job. The hours people currently spend are mostly spent on items that would have been discarded in ten seconds under a written rule.
