Counting Human Rights Violations with GoFundMe

Dan Finlay
Aug 19, 2016 · 4 min read

That’s a great question. My initial google searches revealed what others found, which might seem a little discouraging:

Furthermore, their pagination cuts off at a certain number of pages, showing you the same maxed out results, with unhelpful numbers.

Also, looking at their website’s XHR requests, they’re doing server-side rendering, so we can’t go replicating their own requests to an internal API.

No useful undocumented APIs to speak of.

But I noticed their site does have categories, and the categories have a search bar.

Maybe we can use this search bar to our advantage

At first I was thinking maybe we could try searching for a disease or condition with a known incidence rate that is low enough for the pagination to show an accurate number, and then we could normalize for the rate of other conditions, I found something even better.

Generic search terms FTW!

While the pagination may round up to “thousands”, the search results page will give an exact number! I threw out a generic word like “the” to start, gave me 194k results. Tried another generic term for comparison, how about “a”?

You know it’s counting the records because the page takes a really long time to load!

Slightly different but similar number! I think we have a pretty good ballpark estimate of the number of open issues in the medical category of GoFundMe!

By doing trivial queries on the different categories and totaling them, we can come up with a rate of medical records on the site.

Bonus: Some categories list the amount that they’ve raised, so we can potentially correlate the number of results to an amount of money raised!

I mean, we can correlate assuming people aren’t donating more to Olympic athletes than they are to their neighbors in medical crises. I’m willing to take that leap, that’s reasonable, right?

Ok so I opened the console on GoFundMe, gathered all the links

Spit them into a single text string, line by line:

And I wrote a ruby script to open all the windows so I could avoid typing all the search queries in, and just count myself.

If there were more categories I’d count from the pages automatically too, but at this level it’s probably faster to just look at the pages myself. Also, I usually use node, and I did not try rstrip at first and regretted my 40 tabs of “not a page”.

Anyways, after I tallied them all up, I collected the data in a spreadsheet, you can look at it here.

The results fall in a pie chart like this, making medical expenses a pretty solid quarter of all GoFundMe campaigns:

Which yellow slice is medical, do you think?

Furthermore, “Emergencies” account for another 92,186 search results to “the”, so between Medical and Emergencies, we’re looking at 92,186 + 194,990 = 287,176 medical or emergency campaigns.

If we try to factor in the fact that this isn’t even all of the campaigns, just campaigns with the word “the”, we have to accept that the number is somewhat larger. If we assume the difference is similar to the difference in Olympian campaigns (which have their total campaign count listed), that ratio is 122:140.

If we adjust our medical/emergency number by that factor, we get an estimated 329,546 campaigns.

Stretching the Numbers

Anyways, it turns out only the Olympic fundraiser page had numbers of the quantity of funds raised, $784,453 over their 140 campaigns, which gave us 122 results for “the”.

There are a number of reasons we shouldn’t just extrapolate from this number and take it to mean anything, but for the sake of fun, let’s just make some huge assumptions, like “people fundraise similarly for their ill friends as they do for Olympic hopefuls”, and see how much money in medical funds may have been spent.

We take that $784,453, divide it by the number of results for the (122, we’re normalizing for our input), which gives us $6,429 per result average (slightly below the per campaign average), and then multiply it by the number of medical results for the, (194,990), combined with the “emergency medical” results, which is its own category (92,186), which makes a total of 287,176.

That very rough total is $1,846,254,504, or just under two billion dollars, assuming we fund medical expenses the same way we do Olympians trying to make it to Rio.

Anyways, this is probably less useful as an exercise in how much is being sent to GoFundMe, and more useful as a measure of how many desperately ill people in our society have no recourse but crowdfunding.

Dan Finlay

Written by

Decentralized web developer at ConsenSys working on MetaMask, with a background in comedy, writing, and teaching.