Ask what differentiates digital media from traditional media and the standard answer arrives as a list of eight differences: reach, interactivity, targeting, cost, measurement, speed, credibility, longevity. The list is not wrong, exactly. It is worse than wrong — it is miscategorised. The two labels name distribution technologies, not kinds of media. A newspaper is a newspaper whether it is printed on paper or rendered on a screen; a broadcast is a broadcast whether it arrives by transmitter or by stream. The institutions, the crafts and the judgements are continuous across the labels. What actually changed, when distribution went digital, is a smaller set of structural facts — four of them — and most of the famous eight differences are those four facts seen from different sides. A practitioner who sorts media by era label will misjudge both kinds. A practitioner who sorts them by structure can work either.

The four facts are these. The gatekeeper changed form, not existence. The record’s permanence inverted, and its correctability inverted with it. The counting became exact about transactions and vague about attention. And speed compressed the checking, not just the publishing. The remaining differences — reach, interactivity, targeting, cost, credibility — are real, and they are dealt with after the four, folded back into the structures that explain them.

Fact One: The Gatekeeper Changed Form, Not Existence

The standard claim is that digital media removed the gatekeepers: anyone can publish, so nobody decides. Examined, this is false in both directions. Edited digital outlets are still edited. An online editor’s judgement about what runs, where, and in what words is not a softer judgement than a print editor’s; it is the same institution exercising the same function through a different pipe, staffed by people who can be wrong, can be persuaded and can be held to a standard. For the practitioner it remains what coverage has always been — decided elsewhere, by someone answerable to readers rather than to senders. Self-publishing did not abolish that gate; it built a road around it that leads only to the sender’s own audience, which is a different destination.

What is genuinely new is a second decider with no precedent: the algorithmic feed. A feed is a gatekeeper that is not an institution. It cannot be pitched, argued with, corrected or held to a standard. It has no readers in the moral sense, only metrics. Its criteria are commercial and opaque, in part even to the platform that operates it. The traditional gatekeeper, for all his faults, could be telephoned — could be given a fact, shown an error, persuaded that a story mattered to his readers, in the way the seeker’s discipline has always presumed a person at the other end. The feed cannot. Work aimed at feeds is therefore a different species of work: authored material competing inside a bought-and-earned mixture where the families of tactics blur — and where the labelling discipline matters most precisely because the sorting is least visible. The practitioner’s summary: the gates did not open. One gate kept its keeper. The other got a machine.

The working consequence is that a modern plan contains two kinds of approach that must never be confused in the reporting. Towards the edited outlets, digital or not, the old disciplines apply entire: accuracy, availability, the record in order, the outcome owned by the desk. Towards the feeds, the work is publishing and promotion — authored output, distributed by a system whose favour is rented by quality, by payment or by both, and which owes the sender nothing, not even consistency from one week to the next. A report that merges the two into a single column of ‘coverage’ has hidden the difference between a judgement and a ranking, and the difference is the one this whole comparison turns on.

Fact Two: Permanence Inverted, and Correctability With It

Print’s paradox was that the edition died in a day and its errors stood forever. Yesterday’s newspaper is wrapping by evening, but the copies in the archive, the library and the cuttings file say what they said on the day — uncorrectable in place, correctable only by a later, visible act: the correction, a genre with its own dignity, printed where readers could in principle find it. The record was perishable as a product and permanent as a statement, and everyone in the system knew which was which.

Digital media inverted both halves. The page lives indefinitely — searchable, linkable, screenshotted — and it can be edited quietly at any moment, by anyone with the password, leaving no trace in the text itself. So the digital record is at once more permanent and less trustworthy than print’s, and the correction discipline inverts with it: in print you corrected because you could not edit; online you must correct visibly because you can. The quiet edit is the standing temptation of the digital record, and yielding to it converts an error into a deception. The shelf compounds the problem: an archive or press room page that is silently rewritten stops being a record and becomes a claim about the present wearing the past’s date — which is why the date on the page is part of its testimony. The archive testifies; it testifies to whatever it currently says. Print’s record could not be changed and had to be answered. Digital’s can be changed and must not be. The discipline moved from physics to ethics.

Longevity, the eighth difference on the standard list, belongs inside this fact rather than beside it — and it cuts against the usual telling. Print’s longevity was institutional and accidental: copies survived because libraries, archives and attics kept them, independent of the publisher’s fortunes. Digital longevity is continuous and conditional: the page persists only while the publisher maintains the site, the domain and the link, and it vanishes entire — not faded, not archived by strangers, gone — when any of those lapses. The medium that boasts of remembering everything has built its memory on a subscription. For the practitioner the consequence is practical: the record you rely on being findable in five years is a record you must keep alive yourself, on infrastructure you control, at addresses you do not change.

Fact Three: The Counting Got Exact About the Wrong Things

The standard list presents measurement as digital media’s clear superiority: precise, real-time numbers against traditional media’s estimates. The truth is close to the reverse. Traditional media estimated audiences and knew that was what they were doing. Circulation figures and ratings panels were estimates wearing their nature openly — argued about as estimates, discounted as estimates, never confused with a count of anything. Their vagueness was honest, and the honesty was load-bearing: nobody built a strategy on a ratings figure believing it was a census.

Digital media counts its own events with real precision — impressions served, clicks, plays, completions — and then routinely mislabels the count as the thing it does not measure: attention, engagement, people. A rendered placement is not a mind attending to it; the bought family’s own arithmetic says what an impression is, and it is a delivery event, not a mental one. The precision is real and the interpretation is the fog — a dashboard can report to the unit what the machines did while saying nothing about what any person noticed, believed or remembered. Worse, a measurement culture that cannot tell the difference will optimise for the machines, because the machines are what answer back. The fuller argument about what measurement can honestly claim belongs to the measurement article; here it is enough to state the comparison correctly. Traditional counting was approximate about people. Digital counting is exact about events and sloppy about what the events mean. Exactness about the wrong thing is not an advance in knowledge. It is an advance in confidence, which is a different and more dangerous thing.

Fact Four: Speed Compressed the Checking, Not Just the Publishing

Content speed appears on every list, usually as a marvel: publication in real time against the daily or weekly cycle. Restated as a verification problem, it looks different. The traditional cycle’s slowness was not only a limitation; part of it was a checking technology. The hours before an edition closed were hours in which a claim could be telephoned, a document requested, a second source found. The deadline was a discipline as well as a constraint, and the system’s accuracy was produced partly by that enforced pause — by the simple fact that there was time to be wrong in private before being wrong in print.

Digital speed abolished the pause. Publication is instant, so the checking must now be supplied deliberately — by readiness prepared in advance rather than by the calendar’s accident: the record in order before the event, the facts verified before the announcement exists, the statement drafted to be right at the moment of publication. In a crisis this is the whole game: a statement written to be quietly corrected later is a statement written wrong on purpose, because the quiet edit is not available as a conscience — the screenshot precedes the edit, every time, and the audience for the correction is never the audience for the error. Speed did not remove the checking stage of the work. It moved the stage upstream, into preparation, and it punishes everyone who has not noticed the move.

The checking moved in a second direction too: outward, into the audience. The traditional reader who spotted an error could write a letter and wait; the digital reader checks in public, in real time, in the same thread as the claim itself — and the readership of an expert publication includes the people the story is about, who correct it faster than any desk can. This is usually described as a threat. Structurally it is the old system’s checking function redistributed: verification that once happened in the hours before publication now happens in the minutes after, in front of everyone. Work that was accurate survives it and is strengthened by surviving. Work that relied on nobody checking is exposed at the speed the lists celebrate — which is the same speed, doing the opposite job.

The Remaining Differences, Folded Honestly

Reach and access are real differences, and double-edged in a way the lists rarely note. Digital distribution genuinely lowered the cost of reaching a large audience — and the same pipe carries every error to the same audience at the same speed. Reach amplifies the record, true or false, without discriminating; it is a property of the pipe, not a virtue of the sender. Interactivity is real too, and it is the audience’s action rather than the medium’s property: what audiences do with material — answer it, share it, organise around it, ignore it — is a subject of its own, treated in the article on digital media engagement, and signposted here rather than re-taught.

Targeting and personalisation are real, and they deserve their honest name: they are data practices. The ability to address a message to people selected by their inferred preferences rests on collecting and inferring those preferences at a scale the traditional system never attempted — inferred preferences are surveillance with better marketing — and a practitioner who describes targeting as a pure capability, without its data basis, is describing half a transaction. Cost divides the same way once stated fully: the entry cost of publishing and of buying placement fell, genuinely; the standing cost rose — the cost of being continuously present, continuously publishing and continuously checked in channels that never close. Cheap to start is not cheap to run, and the era comparison that quotes only the entry price is a brochure figure.

Credibility, finally, was the original list’s most balanced entry, and it survives sharpening. Credibility does not attach to eras or to technologies; it attaches to institutions and to their correction records. A traditional masthead publishing online carries its standards with it through the pipe. A digital-native outlet earns trust by exactly the route its predecessors used — by being caught being right, repeatedly, over years, and by correcting visibly when wrong. Audiences understand this in practice even when the commentary does not: they trust names, not formats, and they extend or withdraw that trust on evidence that has nothing to do with whether the name reaches them on paper or on glass.

Work Media by Structure, Not by Era

The practitioner’s summary of the whole comparison is a change of sorting question. Do not ask whether an outlet is digital or traditional; the labels will date, and some already have. Ask the structural questions instead. Who decides what runs — an editor who can be answered, a feed that cannot, or the sender alone? How does the record persist — fixed at publication and correctable only in public, or editable in silence and therefore in need of a visible correction discipline? What do the numbers count — people, estimated honestly, or events, counted exactly and interpreted loosely? And how fast must the checking be — at the calendar’s pace, or in advance, because no pause remains?

Media chosen and worked by their structure behave predictably, whichever technology carries them. Media chosen by their era label keep surprising the practitioner who trusted the label — flattering the new kind for virtues it does not have, dismissing the old kind for limits it no longer faces, and missing the continuities that decide outcomes: that coverage is still a judgement, the record still testifies, the count still measures less than it claims, and accuracy is still the only property that survives every change of pipe. The labels will keep changing. The structures will not.