s09e03: A small collection of things
0.0 Context setting
Tuesday, March 2, 2021.
I am a terrible person because after finishing reading this sentence you will be aware that there is a professional college a capella cover of Moby’s Extreme Ways, which should act as your most recent reminder that if someone can exist, it probably will.
In this episode: influencer cancellation insurance but not really, Microsoft got AI in my Word, nutrition policy and design, More Government Software and The Silicon Valley Ecosystem Continues To Try To Understand Parenting Even Though It Will Never Be One.
1.0 Some things that caught my attention
This episode is mostly smaller things that caught my attention. Let’s go!
Filaxis is a company offering influencer cancellation insurance, only it’s not because it’s a parody (see its terms of service). My gut instinct is that things that get parodied but don’t exist yet have a better-than-average chance of turning into real things, which is a way of making myself feel better about an early January tweet about Lloyd’s of London announcing availability of social media reputation insurance, colloquially known as Main Character Insurance.
Internet friend @eaton had a good thread on this, too, because it’s basically income protection insurance in the same way that I should probably have insurance against my hands suddenly stopping working, preventing me from typing.
Apparently since around November 2019 [Venturebeat], Facebook has been asking (some?) users to upload a video selfie[sic?] as one of the latest salvos in the “prove you’re a real person and not a bot” arms race. This came up again in a tweet just a few days ago on February 26, but predictably I can’t find any reference to it in Facebook’s support documentation. Anyway, that’s not the point. What caught my attention was two things:
a) whether it’s possible to defeat the video selfie test (which apparently doesn’t use facial recognition) by using the trendy new MyHeritage deep nostalgia tool; and
b) the MyHeritage deep nostalgia thing is just the latest example of a non-AI-based company. MyHeritage is a family tree/genealogy website, which is something Americans are totally into, and it’s using deep learning as an adjunct to their primary business. MyHeritage is not a company that is doing deep-learning face replacement as its primary business, unlike Reface, for example. I am sympathetic to this! Someone at MyHeritage has (smartly?) realized that a way to get people interested in their family tree and… heritage… is to bring that heritage to life, and why not use deep learning to do that? My worry, or a consideration I’d have, is that MyHeritage’s tool is another avenue for user-generated content, and clearly there’s no real review mechanism for moderating what users contribute and get animated. (This isn’t the original, but it’ll do: a MyHeritage animated version of a particularly well-known Beyonce photo).
As part of the drive to find new features to include in Word that aren’t, say, non-stupid formatting of bulleted or numbered lists in table cells, Microsoft have brought of something called Microsoft Editor, which is on some level a browser plug-in that helps you write gud, and on another exciting level as shown in some product screenshots, a way for you to make sure that your cv/resume doesn’t include, presumably the wrong kind of:
“sensitive geopolitical references”; or
“perspectives”
Of course I instantly ran my own resume through it and I’m happy to report that my resume does not include any sensitive geopolitical references (or more accurately, that I got a tick mark for that refinement, which I guess could mean that I’ve included the right kind of sensitive geopolitical references), and that I also got a tick mark for vocabulary. My score was 89%!
Anyway, Microsoft Editor’s support document does not tell you how your Sensitive Geopolitical References or Inclusiveness scores (tick mark or… not a tick mark?) are calculated (I mean, I added the word Taiwan and it didn’t immediately flag anything). Opaqueness in this sense is probably bad!
You should probably check out this list (one of many), of Women in AI Ethics.
The FDA has introduced a new Nutrition Facts Label which I excitedly learned about in a promoted tweet. The promoted tweet includes, inexplicably, a fashion runway show where models are dressed up as food items showcasing the nutrition facts label.
This is interesting because… what sort of work and research goes into updating the label? (One answer is: “based on updated scientific information, new nutrition research, and input from the public”)
Some of the visual design has been updated (“The serving size now appears in larger, bold font” and “Calories are now displayed in larger, bolder font”), and some changes were made from a policy point of view, like:
Calories from fat has been removed because research shows the type of fat consumed is more important than the amount; and
Vitamin A and C are no longer required on the label since deficiencies of these vitamins are rare today.
Anyway. There is also an interactive nutrition facts label and the existence of this is also super interesting! It works, in a way, on mobile! I have so many questions! Who is it for? What job does it do? How well does it do its job for those people? The interactive site explains what a carbohydrate is. Should it? Will this site be updated? Should it be? So many questions! Also, because it is a government site, so there is a PDF. (There are actually lots of PDFs).
I mean, should there be an AR app for interactive nutrition facts? Who knows?!
I swear to god if I see a “The New FDA Nutrition Facts Label Made Some Changes, But This Designer Made A Better One For Just $50” in, say, the New York Times, I will lose my shit.
You would probably be outraged that a legacy modernization and replacement project in Arizona for prison inmate management has failed to implement policy changes, which meant that “hundreds of incarcerated people who should be eligible for release are being held in prison because the inmate management software cannot interpret current sentencing laws".
This is, obviously, only one of the latest things to have caught my attention about the state of government, technology and software, so just assume that I have written about all my usual reactions and then add these ones:
You’ve got a bill that was passed in June 2019 changing release credits for inmates, for which someone in the Arizona Department of Corrections said: “We knew from day one this wasn’t going to work… …When they approved that bill, we looked at it and said ‘Oh, s---.’”, so your essay question is: how would you ensure policy like this took into account implementation? (This is, clearly, a long essay question and more imaginary paper for your handwritten answer is available on request; there is no online service to submit your answer)
I mean, this is described in reporting as a bug, but it is also not really a bug in that the software just doesn’t implement the policy change in the first place.
Here’s another essay question: would a business rules engine have made this easier to implement, harder to implement or no change? Why?
The new system is purportedly modular, but “several other programs have failed to perform correctly”, including tracking of inmate health care, headcount, inmate property, commissary and financial accounts, religious affiliation, security classification, and gang affiliations. This is somewhat concerning[sic] because a modular approach to software is, I think, encouraged by the Federal government, and yet it appears none of these modules actually work?
I will be adding this to my increasingly full folder of “horrible things to do with government and its use or lack thereof of technology.”
NFTs (non-fungible tokens) are stupid and interest in them is more about finding new and exciting ways to add scarcity where scarcity didn’t exist, so in that sense the whole horrific mess is less about software and blockchain and so on and more about the creeping (by this point, a fucking enormous tidal wave) of finance eating the world, not software — the software is just helping.
I honestly don’t know what I could realistically write about NFTs that hasn’t already been written, other than the value being writing it in my own words. What’s probably more interesting than the standard response (which I have admittedly run to) of but if you want to keep track of ownership of something then you could just have a database, why can’t you just have a database is that well, that implies that the method of keeping track of something isn’t what’s interesting and the hook to people. Clearly what isn’t the hook is the godawful amount of energy expended in, I don’t know, updating a ledger, in which case the more, uh, manipulative question is: how can you exploit the desire for scarcity for good?
(Maybe exploit wasn’t the best word there).
Lest you think all software is horrible (it is not), some software is good because it saved the lives of two F-16 fighter pilots when they lost consciousness due to excessive g-forces. Yes, I know. The software involved in those planes and the military industrial complex has been directly and indirectly involved in the killing of many, many people.
My snarky observation about Maple, an… app? that has been funded to the tune of $3.5 million to become “the SaaS backoffice for the family” is that sufficiently well-connected individuals who have families are now annoyed about the state of technology involved in managing families and have been able to attract funding, despite for many years now people in families yelling about how horrible it is to, I don’t know, “run the family office” in today’s world. Anyway, “parenting is hard: what if we made it easier?” is both copywriting that I understand and also copywriting that now gets under my fingernails with rusty, er, nails.
While I am fully on board the whole all-calendaring-is-bad-but-especially-for-people-with-multiple-responsibilities-like-families train, there’s just something about the way Maple is put together and the whole gestalt behind it that feels… icky? Perhaps it’s more the feeling that this is yet another attempt to create some sort of zero-sum, winner-takes-all, uh, ecosystem, where Maple ends up being the dominant way of managing families and my worry that it will work pretty well for some families but not others (take a guess at which for the respective categories).
AS AN ASIDE, I’ve joined the ranks of people who manage third party Minecraft servers for bands of roving children and CAN I JUST SAY AGAIN that the whole business of parental controls / family accounts / child accounts continues to be a horrible mess and I have no confidence at all that it will be fixed during my lifetime and if that’s something you don’t want to accept then I am very sorry but it’s true.
That’s it. Thanks as ever for subscribing. How are you?
Dan