Show HN: Isitflatyet.org (Yet Another Coronavirus Map) | Hacker News Reader

Show HN: Isitflatyet.org (Yet Another Coronavirus Map)

isitflatyet.org

5 pointskahdojay7 comments

My wife and I worked on this as our first coding project together! She recently learned how to code and I recently started my software engineering career. It was a good creative outlet for us during these dark times.

We wanted to make a map that didn't feature large overlapping blood red circles everywhere and to give people a better sense of the geospatial data at a glance. We then wanted to show the curves at a more granular level, like the FT is doing so well now (https://www.ft.com/coronavirus-latest), so you can see an entity's curve by clicking on it's marker.

Would love to hear your feedback!

Nice work! I’ve been looking for a visualization like this. My only feedback would be to make the non-cumulative view the default. Given the name of the site, it makes sense to highlight the flattening of the curve.

Thanks, and thanks for the feedback! How would you want to see it highlighted? Show some slope measure for a given curve?

Agreed. "Flattening the curve", at least in its original meaning, has to do with the new case total. Additionally, it is very hard to detect recent changes in a state's results on the cumulative chart.

Oops I misread it as two pieces of feedback even though you explicitly said one. I agree - changed!

The UI is good wrt readability; but the statistics, not so much.

2 concrete improvements would be 1) to change the color coding to cases per population (right now it seems to simply be a hard cutoff of 100 for orange, 1000 for red--which makes comparisons between unequal-sized geographies misleading), and 2) show a smoothed version of the curves (e.g an exponential moving average or somesuch) to handle noisiness in the day to day data.

But, yeah, much easier to read when it isn't overlapping blood red circles.

(Update: typo)

Thanks for the feedback! We did have a lot of discussion about whether we should be showing per capita numbers and throughout the process the numbers have grown quickly and we kept changing it, first from cumulative case numbers, having to change the thresholds, then changing to deaths as a more reliable signal of how bad things are generally, so this is just where we landed as of a couple weeks ago in terms of how to color the markers.

I do agree that it can be misleading to suggest that 1k cumulative deaths in somewhere like Michigan is the "same" (color-wise) as somewhere like Brazil and normalizing by population would address that. OTOH, I do think it's a valid use case to draw the attention to where the absolute death numbers are highest as well regardless of population at least based on our own curiosity, but perhaps that's more appropriately done in a simple ranking table.

We were actually thinking of repurposing the marker colors altogether to reflect on the "flatness" of a geography's death curve, but not sure if we want to trade off on quickly answering the question "who has it worse off, right now?"

Some things for us to consider, thanks again!

Reply on news.ycombinator.com