1wheel

2015 was about the last year you could get away with publishing an interactive graphic with a fixed width — this made it harder do really creative/original work.

https://visxai.io/ has a bunch more too — see the Hall of Fame section at the bottom for some of the highlights.

I also made a dozen of these a couple years ago, my two favorites:

- https://pair.withgoogle.com/explorables/fill-in-the-blank/

- https://pair.withgoogle.com/explorables/grokking/

It can be! Here's a circuit showing how the model processes "PCB tracing stands for" to output "printed":

https://www.neuronpedia.org/gemma-2-2b/graph?slug=pcb-tracin...

Browsers won't use http2 unless https is on — chrome only allows six concurrent requests to the same domain if you're not using https!

There's also https://lmql.ai/

Basically just a bunch of d3 — could be cleaned up significantly, but that's hard to do while iterating and polishing the charts.

I also have a couple of little libraries for things like annotations, interleaving svg/canvas and making d3 a bit less verbose.

- https://github.com/PAIR-code/ai-explorables/tree/master/sour...

- https://1wheel.github.io/swoopy-drag/

- https://github.com/gka/d3-jetpack

- https://roadtolarissa.com/hot-reload/

Here's an example of that with a smaller BERT model: https://pair.withgoogle.com/explorables/fill-in-the-blank/

Mine is 60 lines of js; the markdown library does most of the work.

https://roadtolarissa.com/literate-blogging/

There's a dinosaur version of Anscome's Quartet:

https://www.autodeskresearch.com/publications/samestats

Have you tried plotting the CDFs? Might be easier to read than the overlaid areas.

And if you're using the pattern element, you can drop in any svg element.

https://developer.mozilla.org/en-US/docs/Web/SVG/Element/pat...

The sort order within a category is also wonky, making comparisons hard:

Google: - Health Insurance - On-Site Mother's Room - Fertility Assistance

Citadel: - On-Site Mother's Room - Health Insurance - Fertility Assistance

The NYT uses docs to author interactive articles that don't fit nicely in the CMS.

I've missed that workflow a lot (comments and edits are amazing!); I just open sourced a tiny library that saves a doc as a text file.

https://github.com/1wheel/doc2txt

  npx doc2txt 1StMiAtcY6bY6yEIQp5pVSGdIHSnZG-kFspdmsSzAJdE --outpath gettysburg.txt

How much of an impact does this have?

I prefer the look of things snapped to pixels; I didn't realize it also speed things up!

Nice write up Colin!

fwiw quadtrees don't make much sense when you're showing a tooltip for ~1 million points: https://roadtolarissa.com/scan-sorted/

> another edit: I also believe one reason that this visualization is good is because there's no build process in the JS. The author clearly just edits the code and refreshes the browser. You need that kind of fast edit-view loop to make good visualizations.

You can go a step further and reload code without refreshing the page. This makes it even easier to make the sort of fiddly tweaks that visualizations require while avoid change blindness.

https://roadtolarissa.com/hot-reload

https://github.com/1wheel/hot-server

Could I connect my mbp to this with one cord to get power and external monitors?

> The City paid $30 million for the contract. But the larger issue is that General Electric has already made more than $1 billion dollars selling San Diego residents’ data to Wall Street.

Where are they getting this billion dollars figure from?

I'm working on a web player ripping off the winamp UI.

https://roadtolarissa.com/winampify

A former Al Jazeera America journalist/coder literally just published an A1 NYT investigation.

https://www.nytimes.com/interactive/2019/11/09/us/internet-c...

https://mhkeller.com/

The column version looks like it has less detail along the top of the plot, but I think that's a rendering bug with the moveTo version: there aren't any data points above 1.

I added code for gaps. The tight loop isn't disrupted that much; the number of interruptions is bounded by the x resolution.

added a placeholder issue https://github.com/leeoniya/uPlot/issues/15

Twice as fast! You might be able to get interactive dragging working with this.

https://bl.ocks.org/1wheel/0e7f71ac0325b9be92c9dd19fe4bcc44

Interesting! I wonder if anyone has worked on simplification for when we know x is monotonically increasing.

One really simple idea: group points by their x pixel, connecting the max and min y values for each with a vertical line.

Have you tried running simplifying the lines before rendering? Rendering 150k points is cool, but there aren't enough pixels to see that level of detail.

https://www.jasondavies.com/simplify/

Awesome! How long before this shows up in canary?