← 01 THINKHow does he reason?

Not the person who draws charts

I want to work on showing data as pictures. Wanting is also a claim, so I counted how much evidence there actually is.

This is the entry retold in plain words. The sentences the author wrote are on the other side of the switch.

Search this whole site's code for the letters d3 and exactly one line comes back.

"mermaid bundles d3 and dagre and exceeds 500KB gzipped,"

It only shows up as a reason not to use it.

D3 is a well-known tool for turning data into pictures. And the material laying out the direction I want to go puts it right in the middle of the things to learn.

Wanting to go somewhere is also a claim. This site's rule is that a claim carries its evidence, so I counted the six things to learn, one at a time.

What the count says

Things to learn — evidence on this site, per boxBuilding UI3 entriesDrawing9 entriesThe D3 tool0Handling data2 entriesChoosing wellcannot countUsable by all3 · one missedFour have evidence. One is zero. One I do not know how to count.And the zero is something this site already turned down once, for being heavy.The number is how many entries you can open on this site.

Drawing, 9 entries. Every picture on this site is drawn by hand. No tool. That is why they follow the light and dark screen automatically, and why each one is only a few hundred bytes.

Handling data, 2 entries. One of them counted a team's real records and found that what I thought was a tool for managing tasks was actually a tool for managing releases. 18 releases against 11 tasks.

Usable by all, 3 entries. The brightness difference between text and background is pinned down as a promise, and if it slips the site refuses to be built at all.

D3, zero. And the three things I am supposed to build — a screen that explains data, a map or a time graph, and a reusable set of charts — are all zero too.

The box I could not write zero in

I did not write zero next to choosing well. I wrote "cannot count", because being honest required it.

Drawing nine pictures meant choosing what to show and how, every time. The picture in the entry just before this one draws two boxes deliberately identical. The point of that entry is that the two cannot be told apart on screen, so the picture had to be un-tellable-apart too.

That was a choice, and it was the right one. But I chose it backwards from what the writing needed, not because I knew any theory. I have never learned how colours mislead, or how an axis can be drawn so it fools people. So I have no standard of my own for telling whether my choice was right.

Having zero evidence and having no standard to judge by are different. The first you fix by making something. The second you have to learn first.

The box that hurts most is not D3

The material I read says that when people hire for this, one thing they actually check is "can it be used with a keyboard".

There is one thing on this site worth calling a visualisation: a 3D screen where records stand up as bars. One bar per entry, height is length, across is time order.

And in this site's own notes, I wrote this.

The bars in that view cannot be reached with a keyboard. There is a list as another way in, and putting keyboard access on the bars is worth doing and has not been done.

On the subject I have written about most, the one visualisation I have fails. Brightness pinned as a promise, button sizes raised to the standard — and the bars still cannot be reached by keyboard.

D3 being zero is missing because I never did it. This one is done and then left out, which is a different kind of thing.

Does the reason for turning it down still hold

The reason for not using a drawing tool was this — you cannot make someone download 500KB just to draw one picture. That entry only had 12KB of room.

That judgement is still right. But it does not carry over to D3.

The difference is what the picture is for.

  • On this site, a picture is an illustration helping the writing. Put a big tool behind one illustration and it weighs more than the text.
  • On a screen whose job is showing data, the chart is the product. The scale you weigh it on is a different scale entirely.

I had put two different things on the same scale and expected the same answer. Measuring by weight was right; carrying that measurement to a different problem was not.

Drawing by hand nine times was not wasted either. To know what a tool is doing for you, working it out yourself first is faster.

One thing I was already doing

The last sentence of the material says this.

The goal is not the person who draws charts, but the person who understands what the data means, picks the right way to show it, and builds it into something that works well.

The middle one — picking the right way to show it — is something this site was already doing. Just never with a chart.

This site's mark works that way. It folds the record of the work into a round shape. One slot per piece of work, depth is how many lines that work changed, marks reaching outside are the big stages.

That is data turned into a picture. It just was not called a chart. And nobody can paste that mark onto someone else's site, because their data is different. That is exactly where showing data parts ways with decoration.

What I did not check

Half the evidence here is what I counted myself. The other half is what that material says. The two should not be treated the same.

It names a few real companies as examples. I did not open those job posts myself. I use them only as grounds that such jobs exist, and I am not claiming to have checked what any of them ask for.

If you do not separate what you checked from what you were told, the reader cannot tell which half to believe.

What to do next

  • Do not skip the basics. Nine pictures prove I know the medium, not that I know the theory. That is what "cannot count" meant.
  • Pay the keyboard debt first. The one visualisation I have already fails a standard I already wrote down. That comes before starting three new things.
  • Use D3 first in a product, not in an illustration. Put it in this site's pictures and the reason I turned it down becomes right again.

An unproven box is not the embarrassing part. Not writing down that it is unproven is.