s2e24: The Random Walk Strategy; The Wrong Way Round; Miscellanea 

by danhon

0.0 Sitrep

Then:

Tuesday, August 25, 2015. 36k feet on the way back to Portland from a couple days in the office in San Francisco. I have a bunch of things in my newsletter notes, but they don’t really feel like anything big. At least, they don’t feel like anything big right now – maybe some thing big will fall out of them when I start writing.

So, I guess, a series of random thoughts.

Now: Friday, August 28, 2015. I tried to finish this a few times and then life just got in the way. Apparently life always finds a way.

1.0 The Random Walk Strategy

Small things everywhere that could be better: I’m logged into Aetna, my health insurer, taking a look at my medical claims to date because I’m trying to – well, let’s just say that I’m trying to understand what the hell’s going on with a bunch of bills that have appeared and what I’m supposed to do about them. I can’t do the things that I want to do, namely, get the details about all the claims relating to one particular provider. I have to look at them all individually, and the only way I can filter claims is by family member (ie: me, my wife, or my son). I can’t say: just show me all the ones relating to this provider – which I might want to do because providers bill you separately. Or can bill you separately.

So: I have to click through and open up this provider’s individual claims and then what I actually want to do is save PDFs of them because I don’t have time to do this now, but I do have the time to gather all the records. I can’t command-tab to open them up, because they use some shitty javascript to inappropriately retrieve and refresh the document window using AJAX – which means I have to look at them one by one. And then when I look at each one, I hit “print PDF” which brings up a PDF that I want to save. And then, only when I get to the last one, do I realise this: the regular webpage version of the claim hides information behind a disclosure button – so there’s just one line item and you can hit “more” and it’ll expand the disclosure. Predictably, the information that I want is behind the disclosure. But the PDF that’s generated doesn’t expand the information: it’s just a PDF of the current, unexpanded web view. So I have to go back and do it all over again.

It is things like this that make me want to jack it all in and start fixing healthcare because god knows there’s a lot of healthcare to fix. And then I’m back in my head thinking about the likelihood of success of change in a large existing institution or just starting a brand new one and building from there. So far, it’s not looking great for the existing institutions.

I am excited because Matthew De Abaita’s next book, IF THEN[0], is coming out soon, and his first book, THE RED MEN[1] was a fantastic read about simulated people, simulated cities and, well, a whole bunch of other things. This is me telling everyone I know to go buy IF THEN because I’m sure it’s going to be great.

I just finished reading Warren Ellis’ ELEKTROGRAD: RUSTED BLOOD[1], which Ellis describes as “a crime story set in a strange dream of a possible city.  A science fiction mystery about theoretical architecture, AI and vintage robotics,” and was a quick, easy read and the kind of thing that made me feel a bit useless for not having some sort of architecture-resonant-response inside me because a whole bunch of super smart people I know and like and respect and probably inappropriately idolize are also Know Things About Architecture like walking buildings and Archigram. (Case in point: I didn’t twig that the Archigram I kept reading about in other peoples’ blogs was with a hard C because Architecture. But that might just be me being slow).

After having read William Gibson’s The Peripheral[2], the lingering and dumb idea of looking for signs of intelligence and signals in high frequency trading. Or that there’s someone out there who can do high frequency trading *faster than they should be able to*. I mean, if I were a time-traveller from the future and I happened to bring a computer with me, there’d be a whole bunch of stuff I could do. I mean, what sort of noisy events could you find other signals in?

Greg Borenstein have a podcast-in-progress that, if you’re being uncharitable (and also, I suppose, honest), is a bunch of a middle class educated white-ish dudes, has one episode in the can – more-or-less about the future building in forthcoming Fox TV show Minority Report – but the second episode is probably going to be some reckons around the portrayal of artificial intelligence in movies. The homework for this was for us to both go and watch Ex Machina[3] because neither of us had seen it yet and it was the sort of recent-precipitating-event. In any event, watching *that* movie and briefly chatting about it with Borenstein over Twitter led to the natural (in my head, anyway) jump to a fictional cut of My Dinner with André[4] as a Turing Test (which also led to being sidetracked and starting a new Wikipedia entry, List of Eponymous Tests[5] – which is being proposed for speedy deletion! Please help me by adding eponymously named tests to it!) and then a bunch of tweets about the Bechdel-Wallace test in software (“The Bechdel Software Test is failed when a group of people design a piece of software and no women are involved in it whatsoever”)[6].

(My embarrassing confession about My Dinner with André is that I’m talking about it as if I know anything about what’s clearly a hugely critically acclaimed and revered piece of filmmaking, but the only thing that I really know about is the cultural references *to* it, which means that I’ve watched the Community episode Critical Film Studies[7]). This is, I suppose, how culture works these days.

[0] IF THEN – Matthew De Abaitua, Amazon
[1] ELEKTROGRAD – RUSTED BLOOD – Warren Ellis, Amazon
[2] THE PERIPHERAL – William Gibson, Amazon
[3] Ex Machina (film) – Wikipedia, the free encyclopedia
[4] My Dinner with Andre – Wikipedia, the free encyclopedia
[5] List of eponymous tests – Wikipedia, the free encyclopedia
[6] Dan Hon is typing on Twitter: “The Bechdel Software Test is failed when a group of people design a piece of software and no women are involved in it whatsoever.”
[7] “Critical Film Studies” · Community · TV Review · The A.V. Club

2.0 The Wrong Way Round’

Via Jack Madans the other day in the Code for America Slack, news that the California SAWS Consortium IV[0] – a consortium of counties in California that is working on the statewide automated welfare system (SAWS) – is submitting a grant for money to develop a mobile application. Which, you know, all well and good. It’s good to develop applications. But what’s telling in this case is the language around which the board discusses the reason for developing the application and what the application’s supposed to do. Here’s the relevant excerpt from the minutes of the relevant board meeting[1]:
USDA FNS Process and Technology Improvement Grant Opportunity for Mobile Application

Consortium will be submitting an application for the grant in the following week. C-IV has won this grant multiple times. The goal will be to create a mobile application that will primarily be targeted at contact deflection.

So. I get that these are internal board meeting minutes, but they’re also made available to the public. And as a regular person, if I try to understand what “the goal will be to create a mobile application that will primarily be targeted at contact deflection” means, I *kind of* understand it to mean that they want to create an app that will, well, tell contacts to go away. Because that’s what deflecting means, right?

We could give the consortium the benefit of the doubt and try to understand and read that “contact deflection” actually means “help users self-service or self-qualify so that they can solve problems by themselves without needing to talk to customer service”, but then why not just say that? Why not just say that the goal of the mobile application is to help users solve their problems so they don’t need further assistance from customer service? I mean, saying it *that* way would imply that the consortium were considering things from the user’s point of view and that they saw their job as helping the user accomplish their tasks. But no: this is a stupendously and embarrassingly inward-facing statement that makes it *look* like the consortium wants to not have to deal with its users. Even if that weren’t the case, it’s what it *looks* like. That’s without even digger deeper and saying: well, what do we need a mobile application for? If the sole goal of the mobile application is to “deflect contacts” then does it do anything else? Is it a separate mobile application than the one used to apply for or manage benefits in the first place? Does this mean there’ll be another customer service mobile application whose sole goal is to discourage or stop users from accessing other customer service?

[0] C-IV System Highlights – yeah, don’t talk about the website. It’s not good sport.
[1]  Meeting Minutes – Word document, I’m afraid.

3.0 Miscellanea

With the revelation that most of the Ashley Madison “women” in the database were probably fictional, the rather silly idea that ASHLEYMADISON was an NSA operation that needed a whole bunch of testers in a deniable format for a Turing Test and that Something Happened and the whole thing needed to be shut down in a deniable way. Which is totally unlike the plot to anything I might have watched lately. Of course it’s only a short skip to consider other, hitherto unknown NSA operations such as CHRISTIANMINGLE and FARMERSONLY.

Stupendously smart and dedicated person Matt Patterson[0] has very kindly hand-rolled an Amazon EC2 image[1] that you can just start up on US-WEST-2 that includes everything you need to run char-rnn with my newsletter and 25 megabytes worth of State of the Union addresses. I am looking at you, Robin Sloan, as I believe this is relevant to your interests.

The schedule for next year’s O’Reilly Design Conference is up, which means my session description is up, too. Come and heckle me about it, or send me a note about things you think I should include in the session.

Facebook M is out, which continues the trend of having personal assistant icons/avatar representations be all curvy and windy. Apparently it’s mainly staffed by people, because way more twenty year olds want to work for Facebook than well-trained concierge algorithms.

[0] Matt Patterson (@fidothe) | Twitter
[1] How to build Torch with CUDA extensions with a Ubuntu 14.04 g2.* instance on EC2
[2] The empathy gap: UX, IoT & Interaction Design, O’Reilly Design Conference, San Francisco, CA, January 19 – 22, 2016 San Francisco, CA

OK, thanks! Send notes! All that stuff!

Best,

Dan