The slides who protects

Posted on: 6 June 2026

In 2004 Jeff Bezos prohibited PowerPoint among Amazon's senior team and replaced it with a six-page narrative memo, read in silence at the start of every meeting. The reason he put in writing was not aesthetic. A written narrative, he argued, forces the author to work out what matters more than what, and how the parts hold together, whereas slides grant permission to glide over the ideas and to flatten any sense of relative importance. Translated out of corporate prose, the claim is sharper than it looks. The bullet point is not a neutral way of arranging information. It is a device that erases hierarchy and conceals the holes in an argument. Set three lines one beneath another and they appear to carry equal weight and equal causal force, which they almost never do. Prose obliges you to declare what is cause and what is effect, what bears load and what is mere decoration. The list lets you cheat.

Edward Tufte demonstrated the same mechanism with a case that still chills. In the inquiry into the Columbia shuttle disaster, the information warning of the fatal risk sat buried in a slide, at the foot of a chain of sub-bullets and in a smaller typeface, treated as one detail among many. The shape of the bullet point had hidden the single most important thing by making it look like all the others. This is not an anecdote about NASA. It is what happens every day in hundreds of meeting rooms, where the decision that counts drowns among the housekeeping slides and no one sees it, because the form has granted it the same rank as everything else. I have watched it happen often enough to stop calling it bad luck.

Which brings the question that turns the whole subject over. If the flaw is documented this well, from Bezos to the cognitive research, why does the deck not merely survive but dominate? Because it does not survive in spite of failing to communicate. It survives precisely because it fails. It communicates something else, namely that the person presenting has covered themselves. Sit through enough briefings in Whitehall or pitches in the City and the pattern is unmistakable. The off-the-shelf template is adopted not because anyone has verified that it works but because it removes friction and distributes responsibility. "I did the presentation the way it is done" protects better than any originality, because if the project fails the form will never be laid at the door of the person who chose it. They followed the standard, which belongs to everyone and therefore to no one. Organisational sociologists have a name for this, institutional isomorphism: firms converge on the same practices not because those practices are effective but because adopting them confers legitimacy and shields against judgement. The deck is isomorphism in its purest form. Its flatness is not an execution problem to be fixed with a course in public speaking. It is the outcome the system rewards, because a system that rewards cover produces instruments of cover.

There is a second confusion underneath all this, which is that almost no one separates the two different jobs the slide is forced to do at once. The first is the document: dense, made to be read at one's own pace and annotated. Here the printed sheet you leave in someone's hands at the end of a meeting wins decisively, and the Amazon memo is nothing more than that, the choice of the document over the slide. The second job is live presence, where anything you project that is not a single image or a single word competes with your voice and loses. Richard Mayer spent years measuring this. Putting on screen the same text you are speaking aloud worsens comprehension rather than improving it, because it forces the brain to track two channels saying the same thing slightly out of step, and that lag is noise. The word-heavy slide does not help the listener. It competes with the speaker.

This is where the instinct to dramatise turns out to be right, up to a point. Dramatising not to entertain but to drive a concept into the head and keep it there is the exact opposite of the template. A single strong image fixes an idea far better than an abstract list, because the mind holds on to scenes and lets the lines slide off it, a phenomenon known for decades. A concrete scene or a visual analogy that lands is worth ten summary slides.

The dose is another matter, because before the audience that actually counts, dramatisation carries an asymmetric cost. In front of a decision-maker, past a certain threshold the theatrical stroke does not reinforce the message, it weakens it, because it makes you read as someone selling, and the moment you look like a salesman the content becomes suspect regardless of its worth. Theatre works on those who want to be persuaded and backfires on those whose job is to decide, because the person deciding has been trained to distrust anyone who entertains them. The strong route with that audience is therefore not more spectacle but less. One image that stays, one sentence that will not be forgotten, and then the dense document left on the table so each person can work it through alone. The show that convinces the people at that table is not the one that amuses them. It is the one that lets them believe they understood something on their own.

The proof that all of this is structure and not personal taste is easy to picture, yet almost no one goes looking for it. If format were chosen for measured effectiveness, we would see firms facing the same problem diverge in their formats and experiment, using the document where the document wins and the slide where the slide is needed. We see the opposite: convergence on near-identical templates, indifferent to content and to result. Amazon's experiment has been public for twenty years, it comes from the most data-obsessed company on the planet and it produced visible results, and still almost no one has replicated it. If the choice of format depended on effectiveness, that fact would be inexplicable. It becomes obvious the moment you stop asking whether the deck communicates and start asking who it protects.