Data Glitches and Absurdly Generous Subscribers

This is a special edition of the newsletter because I meant to give the Rodney Smith update on Monday but it slipped my mind. I’m really energized with the giving spirit that has come from this. Every day, people exceed my expectations in kindness and generosity.

• Rodney Smith Jr Matching Update

• Data Glitches and Testing Errors

• Disney Shorts: Gulliver Mickey

Rodney Smith Jr Matching Update

Last week, I introduced Rodney Smith Jr and his Raising Men (and Women) Lawn Care Service and pledged to match up to $1,000 to support his work in encouraging young people to give back to their community by caring for the lawns of the elderly, disabled, single parents, and veterans.

Because you people are amazing, we ended up sending Rodney 9 lawn mowers, 4 weed eaters, and 2 leaf blowers, in addition to almost $1,000 in cash donations. All told, we raised over $3,500 for his work and I wanted to thank everyone who contributed to helping this important effort.

Rodney posted this video which, due to some video time limit, cuts off at 2:20. But you can see the whole thing at his Facebook page.

Rodney Smith Jr @iamrodneysmith

Making a difference one lawn at a time. Full version 2:39.

Data Glitches and Testing Errors

When I was young and silly (well… sillier) I ran a small software application side-business called “Math in Public”. I am apparently enamored with giving the things I create terrible names (apologies to my children MiniMath, MicroMath, and PicoMath) but the idea in my mind was that this name expressed confidence. When we practice, when we’re bad at something, when we’re not ready to show it to the world, we do it in private where no one can see it. Only when we’re prepared and confident are we willing to do that thing out in public.

Since the beginning of this, I’ve tried to communicate exactly how difficult data gathering is and the fact of the matter is that we are all seeing county, state, and federal public health data systems straining under the sudden workload of having to report data every single day when they are accustomed to, at the most frequent, weekly reporting. If you’re really interested in what I think about this topic, I have a piece in the upcoming print edition of Commentary on exactly this topic. Something these health departments always did quietly (real-time data gathering and reporting) is now being done right out in public where anyone can see it.

To a general population who has been conditioned to view data as truth, this makes for a very tenuous situation. Data represents truth, as long as we understand the limitations and boundaries of the data. As long as we know that the process of gathering data is, like all things in our fallen world, imperfect and subject to error.

A couple of recent stories have caught my eye in a way that I want to talk about errors in data. The first one is in California. For someone deep into the weeds of the data, California has been a thorn in our side for some time. Back in April, California has tens of thousands of “pending tests” that just stood there suspended for weeks. At no point in all of this has the COVID count from the state health department matched the sum of the individual counties, often differing by thousands of cases.

The latest weirdness out of California is something that has been raising eyebrows in the COVID data monitoring community for a few weeks. We noticed that California was showing lower cases than expected, diverging from similar regions in a way that seemed odd.

It turns out there was reason for skepticism. California’s disease registry system (CalREDIE) has been plagued with glitches over the last month, losing data submissions without informing the labs that were reporting that their reports were not properly recorded. Eventually, the California Department of Public Health abandoned the system and requested labs try to add their data to an Excel template that could be processed by the state.

And then there was the far less excusable problem where they forgot to renew the security certificate for the website used by one of their most important lab partners.

This is not great news. It seems the problem stems from a software plugin that was meant to accelerate the reporting process but introduced unexpected errors. Welcome to the horrible horrible world of real-time data and software.

This is why I brought up my horribly named ex-business. The California health department is working as hard as they can. Maybe they were sold on a bad product or maybe an installation went wrong and no one caught it. These data professionals are working out in public where everyone can see them. Instead of their mistakes being caught 2 weeks down the line in a code review, they are being caught by state epidemiologists and governors and newspapers and published as scandals.

These are not scandals. This stuff is hard. It is true that most of this work should go through a more rigorous set of testing and quality assurance (QA) before it hits the public view, but that kind of thing takes time. Like… a LOT of time. I’ve seen projects go through 2-3 months of QA before a public release. It’s hard to remember, but three months was a *long* time ago. This newsletter was born three months ago tomorrow. Kind of a lot has happened since then.

California is likely to catch up on their testing errors in the next few weeks. Then we get to try to sift through if we’re seeing actual increases from California or if it’s just the labs and health departments playing catch-up with the data.

Eventually we’ll sort through it. There are some good signs in the immediate data out there… we seem to be coming down from a case high, though we are worried about case increases in new regions (more about this in the next newsletter). I’m a little worried about what this means in the long term, but I do feel confident that the reporting structures and health departments are getting better. Everyone wants to get better. It’s just hard and it takes time.

Disney Shorts: Gulliver Mickey

This is one of the many “Mickey and the Orphans” shorts in which Mickey entertains a dozen or so orphans with his antics or stories or performances. In this short, Mickey calms the orphans causing havoc in his house with a rendition of Gulliver’s Travels in which Mickey plays the role of Gulliver. The main joke, of course, is that Mickey is a mouse and the Lilliputians are mouse-sized people.

Mickey plays it up for full effect, toying with the Lilliputian general for laughs and making a mockery of their army in a game of marbles. Eventually Mickey ends up in battle with a giant spider threatening the town and proves himself a defender of the weak (who he was recently oppressing).

I like this short because it scores on a number of levels. It hits emotionally with the orphans, it holds a sense of the fantastic with Gulliver theme, it has several good visual gags, and it toys with the narrative structure of very old classic films. Ideally, we should see Mickey triumph over the evil spider. Instead we get the thrill of the conflict but the fantasy nature of the narrative allows us to step out of it without resolution. It’s fairly subversive in that way, it gives us 90% of what we expect but surprises us by flipping the script in the final 30 seconds.