Industry
Part of Recording the right date type keeps a policy news timeline useful for years
9 things to weigh up on policy news and timelines risks
Policy news and timelines risks: the failure modes of a policy watch, from alert fatigue and link rot to the single owner who leaves without a handover.
A policy watch fails quietly. Nobody announces that it stopped working, and the record keeps looking like a record long after it stopped being true. These are the ways that happens, roughly in the order they bite.
What to take away
- The most expensive failure is not missing something. It is a record that is confidently wrong.
- Link rot and silent policy revision destroy more records than inattention does.
- A watch with one owner and no handover has a known end date, and nobody has written it down.
Acting too early
Building a process around a proposal is the beginner's failure and it is not cheap. Drafts change substantially during consultation, and a control you built for a text that then changed is worse than no control, because it looks like compliance and is aimed at the wrong thing.
The tell is a project whose justification names no published instrument. If nobody can point at a document with a status, the honest position is that you are preparing for a possibility, which is sometimes the right call and should be recorded as that.
Acting too late because the headline described someone else
The opposite failure, and the more common one. An obligation arrives staged by size, sector or category, and the headline describes the category in ordinary words that did not sound like you. By the time anyone reads the threshold provisions, the date has passed.
The mechanical fix is boring and works: keep "in force" and "applies to us" as separate fields, and never leave the second one blank. Unresolved is an acceptable value. Blank is not.
Alert fatigue
A feed that produces thirty items a week trains its reader to skim, and skimming is how the one item that mattered gets classified as routine. The volume itself is the risk, not the content.
Narrowing the scope feels like accepting more risk and usually reduces it, because attention is the scarce resource. A watch over six named subjects that is actually read beats a watch over sixty that is not.
Link rot and silent revision
Two different problems that produce the same result.
Two Link Rot Problems
Ordinary link rot
- Cause
- Site reorganization
- Result
- Dead or redirected URL
- Fix
- Cite permanent archive
Silent revision
- Cause
- Policy page revised
- Result
- Previous version gone
- Fix
- Copy page when citing
The first is ordinary. Bodies reorganize their sites, documents move, and a URL saved two years ago resolves to a search page or to a redirect that lands somewhere else with a healthy status code. A record full of dead references is not evidence of anything.
Where an official publication system keeps a permanent archive of its own issues, cite into that rather than into a press page. GovInfo's description of what the daily Federal Register holds is one example of a system that does, and the equivalent exists in most jurisdictions.
The second is sharper. A service can revise its own policy page without a changelog, and the previous version simply stops existing publicly. If you need to know what a rule said on a particular date, the only reliable answer is a copy you took at the time.
This is the single habit that separates a usable record from an anecdote. People skip it because taking the copy costs a minute and needing it happens years later. The asymmetry that makes this necessary is set out in the shape of the platform relationship.
The secondary source that became the record
Someone reads a briefing, writes a note in their own words, and the note enters the record without the underlying document. Two years later the note is quoted as though it were the text. Nobody lied and the record is still wrong.
The defense is a tier field on every entry saying where the claim came from, and a rule that an entry sourced only to commentary is marked provisional until someone has read the instrument.
Territory drift
A rule from one jurisdiction is written about everywhere, and coverage in other countries frequently drops the territory. Records inherit the omission. The result is compliance work aimed at a rule that does not reach you, or the assumption that a rule reaching you is foreign because you first read about it abroad.
Where you are established, where your customers are, and where your staff are can be three different answers. More than one set of rules applying at once is normal rather than exceptional.
Systems have their own doctrines for resolving conflicts between their internal layers, and Cornell's account of how one federal system handles a clash between national and sub-national rules shows why the answer is rarely that the stricter rule simply wins. The way obligations actually arrive is set out in the routes regulation takes to reach you.
The single owner
Almost every small watch has one person who understands it. That person leaves, changes role, or has a busy quarter, and the watch stops. Nobody notices for months, because the record still exists and still looks maintained.
Three cheap defenses. Record every pass with a date so a gap is visible on the page. Write down where the sources live rather than keeping the address book in one head. And name a second person who runs one pass a year, purely so that somebody else has done it once.
The register of what can go wrong
| Failure | Early sign | What limits it |
|---|---|---|
| Acting on a proposal | A project with no named instrument | A status field on every entry |
| Missing a staged obligation | Threshold provisions never read | Separate "applies to us" field |
| Alert fatigue | Items marked read in bulk | Narrower scope, fewer sources |
| Link rot | References that no longer resolve | Archive at the moment of recording |
| Silent revision | No dated copies anywhere | Copy the page when you cite it |
| Commentary as record | Entries with no primary document | Source tier on every entry |
| Territory drift | No jurisdiction field | Make it mandatory |
| Single owner | No dated pass log | Log every pass, name a deputy |
| Overconfident phrasing | Present tense throughout | Write the rung, not the conclusion |
The risk that is not about the record
One category of harm goes unaddressed: what happens to individual people when a rule is enforced against them. A watch tells an organization what is coming.
It does nothing for a person whose account is closed on a Tuesday with a notice that names a policy area and explains nothing. Treating the two as the same problem produces bad handling of the second.
What that person can actually do, and what an appeal can and cannot achieve, is a separate subject covered in what happens after an enforcement action.
Common questions
Is it worse to have a poor record or no record?
A poor record, if people rely on it. An empty record makes people check. A stale one makes them stop checking. If a record cannot be maintained, saying so plainly is the safer outcome.
How much archiving is proportionate?
Archive what you would need to defend a decision: the version of any rule you relied on when you decided something, on the date you decided it. That is a small set. Archiving everything is a project that gets abandoned.
We got surprised despite a watch. What is the post-mortem question?
Which rung the thing was on when you could first have seen it, and why the pass did not catch it there. The answer is usually scope or ownership rather than diligence, and both are fixable.
The same discipline applies to reading an enforcement outcome. The useful question is what was established, not what was announced. That distinction is set out in what a resolution does and does not establish.
