s3e28: Something is better than nothing 

by danhon

0.0 Station Ident

Wednesday, 19th October 2016 and, truth be told, I’ve been going through a bit of a difficult patch. Of course, I’m British so when I say “a bit of a difficult patch” you’ll have to use your culture-sensitivity add-on to figure out exactly how bad (or not-bad) things may have been. Suffice to say that at this moment, right now, it makes more sense for me to write some things (however bad those things may be) than to not write anything at all. And let’s see if I can actually hit send this time.

I’m at Ristretto, a small coffee shop within walking distance from home. I’ve had my monthly check-in with my psychiatrist, and today’s exciting news was getting the results from my genetic pharmacodynamic and pharmacokinetic testing[0]! I have been reliably informed that getting such results is “well Gibson” which I suppose is quite cool. I now know[1] that my SLC6A4 serotonin reuptake transporter doesn’t work the same way everyone else’s does, so maybe the SSRIs don’t work that well with me. And I know that my opioid receptor doesn’t work the same well either! How science!

Anyway. I’m at Ristretto. The other day when I was in here there were a couple of hacks from Techcrunch (hack still means “journalist” right?) and I knew this because I could see their screens and they were signed into a Techcrunch slack (although to be fair they could’ve been cosplaying Techcrunch hacks in which case: wow, I’ve discovered a new niche subculture!) and one of them had a Techcrunch sticker on his laptop. They, of course, were two dudes. This threw up a whole fit of association generation in my association generator (sorry, brain) like: why are there two Techcrunch hacks here? Are there lots in Portland now? Why this particular coffee shop? Is there something special about this coffee shop? If there is, is it bad that I live near it? How depressing that the neighborhood I’m living in just happens to be the one with the coffeeshop that the Techcrunch hacks go to! How predictable! This is terrible! Ad infinitum, etc.

Today’s music is 90s-era Madonna because I am easily incepted by environmental factors like: the music playing in the cafe. This means I have already bought Madonna’s I Remember (Theme from With Honors) (1995) and Madonna’s The Power of Good-Bye (1998) and who knows what I’ll buy next.

[0] Genes analyzed in the Genecept Assay test panel – Genomind
[1] I mean, I don’t *really* know, this being the pharmacogenomic industrial complex and all, right? Like, who knows what these results actually mean? Correlate? Cause? So I’ve got lots of reading to do in Google Scholar.

1.0 TKTKTKTKTK

That’s a joke for people who know more than me about editing and journalism and writing. I learned that TK means that something will come later. I have no idea what I’m writing about today, so I put TK a lot. It’s also a signaling mechanism, because I read that we’re all totally obsessed with signaling mechanisms because they’re easier than substantive work. So me putting TKTKTKTK up there and letting you know that I know what it means makes me look smart. I think, though, that generally speaking signaling mechanisms probably work better if you don’t call them out to be shallow signaling mechanisms. (Also, right, it’s easier to just *signal* something than to actually *do the thing* right?)

This is just going to a big LIFO jumble of things because who has the time to deal with Things That Have Caught My Attention when you’re a dad again (thank you) and your preschooler has decided now would be a brilliant time to regress and totally not sleep through the night without constantly waking you up.

Anyway.

* Volvo had an ill-advised competition on social media (see, it’s already horrible) where they asked people to imagine the future of cars and this would’ve gone perfectly fine had I not accidentally seen their promoted tweet (so I think, *in a way* it’s totally their fault because they promoted their tweet and if they hadn’t, I wouldn’t have seen it) so I decided to take them literally and at their word and tweet them suggestions along the lines of “In the future, cars will [blank]” for a chance to win I don’t even know what because honestly if a brand asks me to do some free strategic forecasting for them I will *totally* jump right in because who wouldn’t, especially in public. Anyway, here’s all my entries[0] and here’s the rest of the hashtag[1] which, unfortunately, is me, because I didn’t have anything better to do[2]

* So assume that artificial intelligence (read: lots of “dumb” deep neural nets running on commodity hardware) just become… commoditized because hey it’s just bits running on atoms and we’ve got loads of the former and we can just think up loads more of the former, and before you know it, you probably have things like:

– Hasbro’s first partnership with Amazon, My Little Bot Pony, is the standout xmas toy hit, and comes with a year’s worth of GPU-accelerated learning on AWS’s My Little Pony GPU Farm. After that, of course, you need a monthly subscription to an Amazon/Hasbro GPU farm pasture for your My Little Bot Pony to continue to grow and learn.

– But wait! Some of those incubated little My Little Bot Ponies are going to be abandoned! Because how long does a My Little Bot Pony last, if not until the heat death of the universe if we are to believe Ray Kurzweil and every xmas toy of 2017 achieves a hard-takeoff singularity?! Remember kids: sufficiently deep learning is for life [which in this case, is the heat death of the… right?], not just for a short period of enforced commercialism

– And also remember: sufficiently deep learning is indistinguishable from magic

* But really, though! The cost of spinning up a new “smart-as-a-puppy” instance is going to be so cheap, and hardware, we’re promised, is going to be all over the place. How many more years until deep-learning optimized hardware is more-or-less free and more-or-less everywhere? And then what? You think we’ve got a problem with animal shelters and cute kittens and puppies right now? What about when we’ve got minimally-smart algorithms DESIGNING THE OPTIMAL SMART PUPPY/KITTEN FACE to display to you and whining that they need to be adopted into your home server and embodied into the latest Sphero bot? WHAT THEN?

* And anyway, as conversational interfaces get more prevalent, are we going to just stop being polite? I mean, it’s not as if we’re models of politeness right now, but in principle we’re bringing into being (citation needed and definition required) a whole class of servants that do our bidding (without asking) but are presented with *some* personality and agency (even if they can’t refuse to serve us). Do we have an ethical obligation to be polite to something that pretends to be a human, but doesn’t know or can’t tell (and is ambivalent – for now) on whether we’re being polite to it? BEING RUDE TO ALEXA CONSIDERED HARMFUL turns out to be a prescient and popular Medium article from late 2017.

* Some people threw some deep learning at Asians and discovered that you’re just not trying very hard (ie: as hard as a GPU cluster) if you can’t tell the difference between Chinese, Japanese and Korean individuals[3]

* Oh wait – and this is not a new thought – but if we do end up building bits of intelligent code that are, say, as smart as a puppy, and other people do a good enough job of convincing us that those bits of code deserve to be treated like said puppy, then… a sort of PETA for AI is inevitable, right? And if you do that well I’m not saying that the Republican party is able to hold contradictory thoughts in their platform (they are) but at some point they might need to come out with a position on whether it’s OK to terminate intelligent software (I don’t know… how intelligent is it? Or can we just pause it and promise it we’ll restart it later?)

* All of this is background of course to the most divisive legislation of 2022, signed into law by President Obama (yes: think about it. Can’t be Barack because he already had two terms. What woman could’ve beaten Hillary and denied her her second term?), the BE POLITE (Being Equally Polite to Office and Living Intelligent Talking robots Everywhere) ACT that essentially thought-policed everyone into saying Please and Thank You to all subservient electronic agent intelligent assistants everywhere because no way was the U.S. going to have a do-over on (even just potential) slavery after what it had just started getting out of.

* I continue to have thoughts about digital transformation in government but you wouldn’t know it to look at me.

* The Pentagon just got a massive new telescope and it’s *totally* for just looking at debris and dust and all that junk up there uh-huh there’s no reason at all why they might be interested in taking a REALLY GOOD LOOK at whatever else might be up in GEO[4] – also! This new telescope came straight from DARPA which is practically the real world equivalent of being backed up to the Pentagon in a truck that has STARK INDUSTRIES on the side.

* Diversity means making sure that you have someone who supports a racist, sexist, alleged sex offender on your board because some people in the country are considering voting for a racist, sexist, alleged sex offender.

* These articles about using Lean and Six Sigma to help with your household and your preschooler are not a joke which means… I don’t know what it means. I already make it hard enough for other people to tell when I’m being sarcastic or just making things up[5, 6]. I guess it means I can’t really do my parody, I Ran A Collaborative Design Thinking Workshop With My Preschooler To Ideate New Ways Of Brushing Teeth and submit it to McSweeney’s because real life continues to be fucking irritating.

* Oh right, a whole bunch of tweets happened because my wife got a weight-loss notification after she gave birth because YOU STILL CAN’T TELL WITHINGS IF YOU’RE PREGNANT[7] and in *that* thread, the even more shocking-not-shocking story from @ninavizz that when she interviewed at Fitbit, a UX manager explained (sorry, mansplained) that menstrual and childbirth issues were “too edge-case”[8]

* a joke about how the maths for deep learning neural networks works pretty much just as well with low-precision (e.g. low-precision 8-bit wide fixed precision) maths vs high-precision (32-bit floating point) and how consciousness only spontaneously occurs in high-precision neural networks, resulting in a whole bunch of optimized hardware being kept from achieving sentience and if you want a self-aware bot, you’ve got to spring for the floating-point maths co-processor. Rest easy, your 386/486SX will never talk back to you once more, with feeling.

* Somewhere Amazon has data on how many preschoolers are talking to Alexa. I would love to know what that is.

* Oh right, something about advertising and brands. People love it when I write about advertising and brands. Well let’s just pretend that somewhere, a brand and an agency loved each other very much and then had an idea. And that idea grew and grew and grew until it had a big budget and then when the big budget was ready, it had a production. And when the production was ready, it laid a series of digital video files and those digital files had to nest in a small dark room with snacks and drinks for a couple of weeks until they were ready. After they were ready, some other people had a look at the film and then had an idea for a website that they launched three days later. Nah I’m kidding advertising is totally different these days. Oh and then someone a hashtag.

[0] hondanhon #futurevolvocontest – Twitter Search
[1] #FutureVolvoContest hashtag on Twitter
[2] No wait! I tell a lie! I totally had better things to do! What is wrong with me!
[3] Title: Do They All Look the Same? Deciphering Chinese, Japanese and Koreans by Fine-Grained Deep Learning
[4] The Pentagon’s massive new telescope is designed to track space junk and watch out for killer asteroids – The Washington Post
[5] How To Use Lean Manufacturing And Six Sigma To Run Your Household | Fatherly
[6] This Time Management Strategy Took An Hour Out Of My Son’s Morning Routine | Fatherly
[7] Dan Hon on Twitter: “#tfw there’s no way to tell your smart scale you were pregnant, or that you’re post-partum now because why would women use these things https://t.co/gMbEy093aM”
[8] nina eleanor alter on Twitter: “@hondanhon The UX mgr @fitbit srsly mansplained to me how menstrual & childbirth issues were too “edge-case” to bthr with. @sara_ann_marie”

OK, enough. That’s it. Bye again for an undetermined period of time while I get my head-shit sorted out.

Well, was it? Was something better than nothing? I don’t fucking know. You tell me.

Best,

Dan