Loading today’s stories
Build1 distinct publisher2 min readUpdated
A dev.to walkthrough, checked in Excel rather than recalled, finds pasting both bypasses a validation rule and removes it. That changes what Circle Invalid Data can still find.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
A validated column is a tally rather than a state. Some of its cells hold the rule, the rest do not, and the grid draws both the same way. The fix for that is ordering rather than a better tool: re-select the whole data range, apply the rule again so every cell carries one, and only then circle [3]. Run it in the other order, as the sequence in Nocito's August 8 page implies most people will [1], and the column that comes back cleanest is the one that has been pasted into most.
The list source has the same shape of problem underneath it. Excel refused the structured reference and accepted it wrapped in INDIRECT [7], which means the product will take the reference as a quoted string after declining it as a reference. Worth fighting for, because a table grows when somebody adds an option and the dropdown grows with it [13]. A defined name set to the same table reference gets you there with a Source box a colleague can read [8]. A defined name over an ordinary fixed range is the version that quietly stops working, because the range does not grow when the options do [9].
Then keep the list off the data sheet. The argument Nocito makes is operational rather than tidy: a list parked beside the data gets sorted, filtered, cut and pasted along with the rows around it [12], and a dropdown whose source was re-sorted underneath it fails as a cell that accepts the wrong value rather than as an error message. Nothing in the workbook points at the cause.
All three of these land in the same place. The rule is an attribute of a cell that any ordinary editing action can remove, so the only durable statement about a column is the one you make by re-applying and re-auditing it, not the one you made the day you built the dropdown.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
The article is by Michael Nocito, described as a data analyst, published on dev.to on August 8, 2026, and is presented as about twenty-five minutes of work with a dropdown deployable in about a minute.
Nocito states that validation checks a value at the moment somebody types it into the cell: it is not a property of the data, it is a gate on one entry route, and there is more than one route in.
The page's diagram shows two routes approaching a cell: the upper route labelled typed meets a solid section of barrier and turns back on itself, while the lower route labelled pasted meets a broken dashed section with a gap and passes straight through into the cell.
In the diagram, beyond the gap the fragments of the broken barrier are drawn drifting apart, showing that the pasted value did not merely get past the gate, it took the gate with it.
Nocito says the behaviour claims were checked by building rules in Excel and inspecting the workbook afterwards, so what is reported about pasting, about table columns as sources, and about which cells still carry a rule comes from reading the workbook rather than from memory.
Nocito instructs readers to run Data, Data Validation, Circle Invalid Data on the column after applying a rule, because rules only govern what happens next and the existing data was entered before the rule existed.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Self-verified single source, concrete but unscoped
The behavioural core is unusually specific for a tutorial: a named test with a before/after tally (three validated cells, two still carrying a rule after pasting two different ways), an explicit accept/reject pair for the Source reference, and a stated method of inspecting the workbook rather than recalling behaviour. That lifts it well above assertion. It is capped, though, by having exactly one publisher, no Excel version, build or platform named, no vendor documentation or independent reproduction cited, and no artefact a reader could re-run.
No adoption signal supplied
The cluster contains no release, deployment, benchmark, pricing, licensing or usage disclosure. Nothing indicates how many workbooks, teams or organisations are affected by the paste-strips-validation behaviour, or how widely the recommended patterns are used. Excel's general ubiquity is not evidenced by the supplied material and is not inferred here.
Slightly overstated generality
Framing and evidence are close to aligned: the dramatic line -- paste does not merely bypass the gate, it takes the gate with it -- is exactly what the reported test showed, and the author explicitly declines to call it a bug, explaining it as ordinary paste semantics. The small positive gap comes from scope, not from spin: a three-cell test in an unnamed Excel version and platform is generalised into an unconditional statement about how pasting behaves, and the diagram's drifting-fragments imagery amplifies a mundane property-copy mechanism.
Practitioner audience-building, no disclosed commercial stake
The visible incentive is an individual data analyst publishing a tutorial on a developer blogging platform: reputational and audience-building, with a mild pull toward a memorable failure-mode framing to make a routine spreadsheet topic land. No vendor sponsorship, product, affiliate offer or paid tool is present in the supplied text, and no claim is made that would benefit a named commercial party, so distortion pressure reads low.
Plausible and internally consistent, unreplicated
Confidence is moderate. The mechanism is coherent, the claims are cheap for any reader to check, the stakes are low, and the author states the behaviour was read out of a workbook rather than recalled. Against that: one publisher, no independent replication, no environment scoping, an internal date inconsistency between the byline and the source metadata, and zero adoption evidence, so the derived operational consequences remain reasoning rather than observation.
build
The query ran and the number was 38 percent high: reviewing AI-written SQL1 distinct publisher
build
Excel Guessed the Alphabet, and 4,685 Names Went Wrong Without an Error1 distinct publisher
build
Three manual interventions in a month, and every guard was working as designed1 distinct publisher
build
Six MariaDB versions, one real difference: the only reason to leave 10.6 is the July 2026 clock1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 23, 2026