中文
Figma Config

A former Stripe designer answers his boss's "why can't we look more like Bloomberg?" with a five-rung ladder of density

Dense by design ft. Matthew Ström-Awn (mu.design) | Config 2026 · Matthew Ström-Awn

22 min
Design-to-Code

22 min total·Actually worth watching closely: ~14 min·3 must-watch clips

Orange = the 14 minutes worth watchingFor the rest, the guide is enough
Segment guide · 8 segments
  1. 0:11 3:34Watch

    One question, a year of thinking

    The speaker recalls presenting at Stripe when Patrick Collison put the Bloomberg terminal on screen and asked why they couldn't do more of that. The question challenged the team's whole value system around white space and hierarchy, and became the starting point for everything that followed.

    Valuable is not the same as beautiful — one of the ugliest pieces of software around is also one of the most valuable in the world.

    The opening puts both screenshots side by side; the wall-to-wall financial terminal against the clean dashboard is the contrast the rest of the talk depends on.▶ Jump to 0:11
    Speaker · Matthew Ström-Awn
  2. 3:37 6:10Skim

    Rung one: visual density isn't the answer

    Takes the most intuitive reading of "density" — how much is crammed onto the screen — and tests it. The speaker shows examples that are beautiful (McMaster-Carr) and examples that are genuinely horrendous.

    Both are packed with stuff, one is a joy and one is a disaster, so how busy a screen looks tells you nothing about whether the design is good.

    Mostly a run of website screenshots as good and bad counterparts; getting the rough shape of each page is enough, no need to follow every line.▶ Jump to 3:37
    Speaker · Matthew Ström-Awn
  3. 6:13 10:11Watch

    Rung two: information density and the data-ink ratio

    Introduces Tufte's data-to-ink ratio: strip out the decorative visual rhetoric, let the data speak for itself, and communicate more with fewer pixels. The counter-example is a chart crime that trades accuracy for emotional effect. The finale is the 1885 Paris–Lyon train timetable.

    Information density isn't about how crowded the picture looks but how much real data each bit of ink buys you.

    The persuasion here is entirely visual: the over-decorated chart and that century-old timetable have to be seen — you need to watch how the lines encode speed and stops.▶ Jump to 6:13
    Speaker · Matthew Ström-Awn
  4. 10:16 12:35Watch

    Rung three: arrangement itself speaks

    Shows the same screen with nothing added and nothing labelled, only rearranged — and suddenly people read it faster and far more precisely. Then a systematic pass through the Gestalt principles: proximity, similarity, common region.

    Without adding a single word, position alone changes how fast and how accurately people understand.

    The before-and-after screens are the entire evidence for this claim; only by seeing the same content rearranged do you believe "nothing was added."▶ Jump to 10:16
    Speaker · Matthew Ström-Awn
  5. 12:35 15:12Listen

    Rung four: the ceiling on meaning, and time

    Points out that compression has a limit: past the brain's absorbing capacity, people get nothing at all. That pivots the talk to time, with waiting rules bucketed by perceptual thresholds.

    Under 100ms, do it instantly with no animation; 100ms to 10s, transitions and a loading indicator; 10s to a minute, a progress bar; over a minute, let people leave and notify them when it's done.

    These thresholds are actionable numbers given purely out loud — take notes as you listen, there's nothing on screen you need to watch.▶ Jump to 12:35
    Speaker · Matthew Ström-Awn
  6. 15:15 17:40Watch

    The top rung: the value density formula

    "Where's Waldo?" shows a design that maxes out visual, information and meaning density while still being terrible in time; subtracting density then makes the same task far faster, and it all resolves into the talk's core formula.

    Value density is the value someone gets divided by the time and space you ask for in return — value per second, value per pixel.

    You need the packed search image and the pared-back version together to feel how much time was actually saved; without that the formula stays abstract.▶ Jump to 15:15
    Speaker · Matthew Ström-Awn
  7. 17:40 19:33Watch

    Density depends on who's looking

    An ECG goes up on screen: to most people it's a smear of spikes, to a doctor it's the difference between saving a life or not. The density of the same design shifts with the viewer's expertise.

    Density isn't a fixed property of a design — it's a relationship between the design and the person looking at it, so talking about density apart from users is meaningless.

    The trace is right there, and your own layman's reaction to it is the argument; turn it into text and the effect disappears.▶ Jump to 17:40
    Speaker · Matthew Ström-Awn
  8. 19:37 22:12Listen

    Less talking about craft, more craft

    The speaker admits everything he shared is old news and other people's work — Tufte, the Gestalt psychologists, decades of HCI research. The real problem is that we talk about craft too much and go do it too little. Then he returns to the question that started it all.

    A good question's worth isn't whether it was right, but how long it keeps you exploring.

    The close is pure argument and self-reflection with nothing to look at — fine to listen to on your commute.▶ Jump to 19:37
    Speaker · Matthew Ström-Awn