Every agency I have worked with has lived through at least one moment when the ground was said to move. Desktop publishing, the commercial web, social, mobile, now AI — each arrived with the same announcement: this changes everything. And each time, two camps formed almost immediately. One insisted that nothing fundamental had altered and the firm should hold its nerve. The other insisted that everything had, and the firm should burn the old model to the ground.
Both camps were usually wrong, and they were wrong in the same way: they had no method for telling the rate and kind of a change from the noise around it. That is precisely the work that thinking historically does. The historian's question about any apparent rupture is not "is this big?" — everything feels big from inside it — but "what kind of change is this, and how would I know?"
The tool that becomes the water
Carlota Perez has the most useful frame I know for this. In her account, capitalism does not advance smoothly but in great surges, each set off by a technological revolution — and each running through the same shape. There is an installation period, finance-led, turbulent and unequal, when the new technology is exciting, scarce and overvalued; a crash; and then, if institutions adapt, a deployment period, when the technology stops being a marvel and becomes the quiet infrastructure of ordinary life.
The detail that matters for an agency is what happens to the tool itself across that arc. Early on it is a disruption — a capability so scarce that holding it is itself a business. Later it becomes what Perez calls common sense: embedded in how everyone works, no longer a source of advantage but a condition of staying in the room. The web was a premium once. So was social. The capability you once charged a premium to hold becomes, soon enough, simply the price of being in the room.
This is the first discipline change-over-time imposes. The right question about a new tool is almost never "is it good?" It is where are we in its cycle? — because the same technology pays completely differently depending on the answer. Mistake a deployment-phase commodity for an installation-phase disruption and you will price a service nobody will pay a premium for. Mistake the reverse and you will sit out the years when the margin was actually there.
The pattern, and the rate
History does not move at a constant speed. Francis Gavin borrows a term from evolutionary biology — punctuated equilibrium — to describe what he sees in the record: long stretches of relative stability broken by short bursts of rapid change. The mistake is to read the velocity of the present as the permanent state of things. The frenzy of an installation period feels like the new normal; it is, almost by definition, temporary.
Set against this, the comforting line that "we have seen all this before" is only half a discipline. Margaret MacMillan is sharp on the danger here: humans cannot help reasoning by analogy, and the analogy we reach for is rarely the apt one — it is the flattering one, the precedent that tells us we already know what to do. "We survived the web, we'll survive this" is exactly that kind of analogy. It may be right. But it is doing the work of reassurance, not of analysis, and the two should never be confused.
So the historian's move is to insist on the disanalogy alongside the analogy — to write both columns, as Neustadt and May taught their students to. Yes, AI rhymes with the earlier waves: a scarce capability, a gold rush, a wave of overclaiming, a coming commoditisation. But name, precisely, what is not like the last time. Each previous wave mostly changed how fast the existing work could be done. The honest question about this one is whether it changes who, or what, does the work at all — and that is a difference in kind, not degree. Where exactly the current wave falls on that line is the question worth sitting with — and the one the announcements are least willing to answer honestly.
The trap of the unchanged process
There is a quieter failure that change-over-time exposes, and it is the opposite of panic: the failure to notice that conditions have changed — often for the better — while your processes go on defending against a danger that has passed. Gavin's checklist puts the question directly — how are things trending? — precisely because organisations so reliably miss the moment when the threat they were built to answer has quietly receded. The firm that survives a disruption frequently freezes the habits that got it through, and then runs those habits for a decade after the threat they answered has gone. The crisis playbook becomes the operating manual. This is change over time read backwards: mistaking the last war for the permanent condition.
Taleb supplies the corrective at the other end. His warning against neomania — the reflex to assume the new thing is the better thing — is the necessary balance to all of the above. Against it he sets the Lindy effect: for ideas, methods and institutions, the longer something has already survived, the longer it is likely to last. The craft skills, the relationships, the editorial judgment that have endured through three disruptions are not obsolete simply because a fourth has arrived. They are, on the evidence of their own survival, the parts most likely to still be standing when this wave has settled into common sense like the others.
What it changes on Monday
None of this resolves into a prediction, and it is not meant to. The use of thinking historically here is to replace a binary — hold the line or burn it down — with a better set of questions. When the next tool is announced as the end of the world as you knew it, three are worth asking before you reorganise anything.
Where are we in the cycle? Is this a scarce disruption you can charge a premium to hold, or a commodity you simply have to have? The honest answer is usually "earlier than the sceptics think and later than the evangelists claim."
What kind of change is this — rate, or kind? Is the tool making the existing work faster, or changing what the work is? Write the disanalogy, not just the analogy.
What of ours is Lindy? Which capabilities have already survived every previous wave, and are therefore the ones least safe to throw overboard in the rush to the new?
The caveat is the one Taleb and MacMillan between them insist on, and it cuts against the comfort the Lindy effect offers. Sometimes a change really is different in kind. Some firms genuinely should have burned the model down and did not, and the survival record of their old methods was exactly the story that reassured them on the way down. The point of knowing the history of disruption is not to be soothed by it. It is to be able to tell, this time, which kind of moment you are actually in.
So the question is not whether this one changes everything. It is the harder one the evangelists and the sceptics both dodge: which everything — and how, precisely, would you know you were wrong?
This is a companion piece to the argument at the argument. If it resonates — or if you want to push back on it — I'd be glad to hear from you: chris@christhurling.com.