# The slowest link on the chart isn't the machine

> A dot-matrix latency chart lists EYE, HAND and LEG in single digits, then RECALL at 240 — the one bar a firmware update was never going to fix.

August 19, 2026 · https://justog.club/stories/neural-latency-chart

A dot-matrix chart, pinned flat like it came off someone's wall. Four horizontal bars, labelled left
and figured right: `EYE 12`, `HAND 31`, `LEG 88`, `RECALL 240`. The last bar runs far past the other
three and carries the only red ink on the sheet.

Three of those numbers describe hardware. The fourth describes a person, and nobody printed a fifth
bar for the upgrade that would fix it, because there isn't one.

## What is this chart actually measuring?

Reaction time, broken down by which part of the system the delay happens in — three augmented and
one not.

`EYE`, `HAND` and `LEG` are milliseconds through cybernetic pathways: signal in, signal processed,
signal acted on. Small numbers, because that's what the augmentation was built to deliver. `RECALL`
measures something upstream of all three — how long it takes to remember what to do before any of
the fast parts get to act on it — and it's the one figure the hardware never touches.

## Why is RECALL so much slower than everything else on the chart?

Because it isn't hardware. It's memory, and memory doesn't get faster because the limbs around it
did.

An eye, a hand and a leg can be rebuilt with faster signal paths — that's a wiring problem, and
wiring problems have engineering solutions. Recall is retrieval: finding the right response among
everything a person has ever learned, under pressure, before the fast parts downstream have anything
to act on. Nothing about improving reaction time in a hand changes how long that search takes. The
chart makes the gap visible by putting both kinds of number on the same axis, as if they were the
same kind of thing.

### That's the chart's whole argument, made without a caption

Nobody wrote "the bottleneck is human" anywhere on the sheet. They didn't need to — four bars on one
scale does the arguing. Whoever printed this most likely built it to justify the next round of
limb work, the same way [a compatibility matrix looks like engineering while doing a different job
entirely](/stories/augment-compatibility-chart). It ended up documenting something the
budget wasn't asking about.

## What makes a good version of this chart?

Three fast bars close enough in length to read as a cluster, and one long bar that has to dominate
the sheet without needing an explanation printed next to it.

Four rows is the right count — enough to establish a pattern with the first three before the fourth
breaks it. If every bar were roughly the same length there'd be no chart, just a list. The whole
design depends on `RECALL` being disproportionate enough that a reader clocks it before they've read
a single label.

## Whose numbers are these?

Nobody real's. No specific person, augmentation brand or clinical measurement is being claimed —
it's an invented reading off an invented body, and the specificity that makes it convincing is
exactly the kind that doesn't need a real source behind it. A genuine clinical chart, with a real
patient's data on it, would be an entirely different and much more serious thing to put on a shirt.

## Does the same chart work for a body that isn't augmented at all?

The shape of the argument doesn't need any hardware in it.

Any set of measurements that improves everything upstream of judgement and nothing downstream of it
makes the same point — faster tools, same slow decision. A racing driver's reaction times against
their read of a changing track. A surgeon's instrument precision against the pause before choosing
which vessel to clamp first. The augmented limbs are one way to draw it; the gap between fast parts
and slow judgement is older than any of this technology.

## Isn't a chart like this just an argument for better augments?

That's clearly what it was made for. It's also, by accident, an argument against the premise it was
built on.

Whoever commissioned this chart wanted a case for the next upgrade — three good numbers, one bad
one, obviously the bad one needs work. Read straight, though, it shows the opposite: the three
numbers that keep improving aren't the problem, and the one that is has never moved. A chart
designed to justify more hardware ends up being the clearest evidence that hardware was never where
the delay lived.

## Does it print well?

Yes, kept to two inks and a strict bar-chart grid.

Charcoal carries the three short bars and the axis labels; red is reserved entirely for `RECALL`, so
the eye finds it before reading a single number. [A limited, deliberate ink assignment reads clearly
the same way a colour-coded loom does](/stories/wiring-loom-colour-code) — one colour per
category of information, nothing decorative added on top. Dot-matrix texture on bone cotton wants
flat bars with square ends, not a gradient fill, or the length comparison the whole chart depends on
goes soft.

## How do I get one made?

Describe the chart at [JustOG](/) — which four figures, and which one runs long. Pick a direction,
drag the crop frame, see it composited on the real garment, and it's made to order and shipped.

Designs other people have published are in the [shop](/shop).

Three bars built for speed and one that was never going to move, pinned flat on the same sheet.