s06e03: Sherlock Holmes and the Slack Channel Names
0.0 Station Ident
This episode’s title is a reference to a tweet I made (back oh so many days ago when I tweeted, etc.) where I thought-out-loud: “Sherlock Holmes is given a list of your company’s Slack channel names and instantly diagnoses the top 10 issues, and which of the leadership team will quit in the next 6 months and in what order.”
1.0 Some of the things that caught my attention
1.1 Graphene in things
Last episode, one of the things that had caught my attention was the discovery that salt-infused graphene makes a good infrared cloaking device, and now apparently graphene is popping up everywhere in my hallucinated sensorium, because via Ahmet A. Sabanci, news of a graphene-infused jacket from Vollebak [Fast Company, Vollebak]. Apparently the jacket uses graphene nanoplatelets (which itself is evocative of a sort of dull, gunmetal-gray extremely militaristic version of your haemoglobin but… probably really bad at what haemoglobin does?), and *one* of the things that caught my attention FastCo’s writeup was this aside:
“The hope is that once people get their hands on the [$695] jacket, they’ll start experimenting with it–like beta testers for the product and researchers for new applications. “[W]e’re looking to harness the collective power of early adopters as a test group to do it,” Tidball says. “[We] believe that between them they’re likely to discover things that we simply don’t know. It brings massive scale to our ability to experiment with the material to see what it can do.”
1.2 Technology projects considered harmful
My most recent trawl of IEEE Spectrum turned up the latest entry in Risk Factor, IEEE’s risk analysis blog, which is happily right next to the Risks Digest in my head (a “Forum on Risks to the PUblic in Computers and Related Systems”, and part of the ACM Committee on Computers and Public Policy).
Anyway, from IEEE’s Risk Factors, we learn about two messed up technology projects - that Natonal Grid (not the UK entity, instead the Northeast US utility company) settled with IT vendor WiPro and that IBM and Bridgestone had a bit of a tiff over a botched SAP-based ERP implementation.
There’s a few things interesting - well, somewhat interesting at least - from the accounts of these two failed projects. The first is that both the customers and the vendors appear to be at fault. In National Grid’s case, an audit report found that a) the vendor didn’t have the experience the claimed to have; b) “the development team [failed] to communciate with the users of the system to find out if it worked” (it will be most amusing to see if anyone expected any “agile” software development to have been happening); c) the customer “[underestimated] the complexity and risks involved in the ERP system development”; and d) the “lack of effective governance and oversight of the project at all levels of the organization”.
The article ends saying that “each party probably got what it deserved” - on the customer end, they got IT systems that cost them lots of time and money to fix (though to be honest, I’d hazard a guess that the systems probably still don’t do what they were supposed to do in the first place, and may not even do what the systems they were supposed to replace, either) and that Wipro and IBM have “faced significant knocks to their reputations for not being able to deliver what they promised.” But if one of the problems is that customers don’t even know enough (or, even, are so incompetent) to check if vendors have the experience they say they have, then there’ll be no shortage of clients for Wipro and IBM in the future.
1.3 But presumably some technology projects still work?
That said, via both Tim Maughan and Jay Owens, Miriam Posner’s See No Evil from issue 4 of the always execllent Logic Magazine points out that 2018’s global supply chain is completely reliant upon software. The failed (sorry, not-yet-succeeded) technology projects above were both ERP projects - enterprise resource planning - using SAP, pretty much the equivalent of the MIcrosoft Office of managing your supply chain in, I guess, a potentially very expensive and broken way that results in suing your IT vendor.
(At this point I have an aside where I have to admit how this is all very peculiar: my dad, you see, is a professor of manufacturing engineering and while I was growing up he’d have all this stuff about what Toyota was doing in his study and every so often he’d ask me to help him with the layout of a grant proposal or paper submission in Word and I’d see phrases like “supply chain management”.)
So what’s interesting (well, not really that interesting, I suppose) is the tension between horrific software implementations that parties think are worth going to court over (mainly, I think, to protect their reputation? Because clearly on the facts, both parties in both cases above don’t come out well) and, well, software that works well enough to be the backbone of a global supply chain that reliably makes sure however many million people get iPhones when they want them (and, to be fair, food and petrol and other goods).
I mean, imagine if we were *good* at software projects?
1.4 A Speculative Future Aside
Tim Maughan, by the way, has by all accounts an excellent book coming out next year on a post-internet world that’s probably got minmial privacy and reeling from whatever self-inflicted harms result from a glut of fake news, so if you’re into that kind of thing you should probably pre-order Infinite Detail. You may also remember Tim from an excellent piece of horrifying fiction about zero-hours contract workers in 2023 and the merger of Foxconn factories with the American prison-industrial complex.
If you liked those two, then you’d probably also like (according to my bayesian model of what you like, based on what I like) Madeline Ashby’s Tierra y libertad about an investigator tracking down emergent sentiences when complex environments stuffed full of complex software mean that kind of thing happens pretty often.
1.5 Fraa Musk and the Cycling Hemmas
* that thing about using pumped water to store energy generated (captured?) by renewable sources so they can be used later, when it’s more convenient to humans? The idea being: when electricity’s cheap because of a glut of wind or solar or whatever, use it to pump water up a big hill, loading the water with a bunch of potential energy - you remember your high school physics lessons, right? Well that’s all very well but you see water isn’t particularly dense and you know what *is* dense and we might have lying around, or have the infrastructure to make lots of? Concrete. So why not use (electric) cranes to lift concrete blocks when you’ve got cheap electricity? Better still, why not automate all of it?
Of course if you do this, then you’ve inadvertently found yourself in a science fiction/fantasy story set in the past (TWIST! IT’S ACTUALLY OUR FUTURE!) and the Apts in the Arbre are talking about maintaining the Hemmas to make sure the Baskip Carriers keep moving up and down, storing and releasing Mir energy, leading you, eagle-eyed reader, to shout excitedly: THIS IS A CRANE-BASED CONCRETE BATTERY SYSTEM!
2.0 And finally…
* As London Gatwick turns to whiteboards to display flight information when display screens fail, operations managers inadvertently invent a new Redundant Array of Inexpensive Dry-erase markers.
* Steven Frank discovers an attempt to out-do those Amazon reviews that say “I don’t know, why are you asking me” in response to questions like “Was this product easy to use?” where Yelp reviewers leave Yelp reviews as reviews for the app, not… reviews for the business.
* via Nelson Minar, a commercial lifeguard drone rescued a swimmer in Sagunto, Spain (New Atlas, The Drive). The drone is decked out in the traditional lifeguard garb of red and yellow with RESCUE written on it and if you give it a couple of years, lifeguard drones will probably start looking like Chris Foss-ish Thunderbird 2s.
* via Hacker News (sorry, but at least I’m being honest), Anthony Madrid wrote about pop songs, sung in English, that contain mistranslated idioms. Example: Ace of Base’s All That She Wants (sorry not sorry for the earworm) where the chorus goes “All that she wants / is another baby / she’s gone tomorrow, boy” and which on closer examination, the intended meaning was “All that she wants / is another *lover*”, i.e., not another baby.
* Facebook now pays about $17 million dollars a year to protect Mark Zuckerberg. As Bloomberg points out, Zuckerberg is worth $68.6 billion dollars on the publicly traded index of “how much an individual human being is worth”.
* If I’m reading things right, one of the reasons why raytracing is suddenly viable as a realtime-ish rendering technique is… because of deep learning? I discovered Matt Pharr’s blog a few weeks ago and his post on why he’s heading to nVidia recaps that one of the big advances is of using deep convolutional nets to do image de-noising. So, where we might have needed to, I don’t know, thousands of rays-per-pixel to generate high quality images, now we need… significantly fewer? There’s something really itching at my brain about this - the idea that we’re, at a high level, training deep neural nets on what makes images look real (ie: like they’re the result of an unreasonable number of physical and optical interactions) and then applying that to lower-fidelity images. I’m struggling with finding the right kind of analogy here - it feels a bit like it’s adjacent to some of the top-down predictive processing where I can imagine if you’d be able to extract some sort of signal from a low quality image (hey, this looks like it should be transparent, or whatever), then here’s a *prediction* about how light rays might interact with it, *without having to do the calculations involved*, and then that the prediction is, from our point of view, good enough. Anyway, here’s a SIGGRAPH talk that Pharr links to from 2017 on how deep learning is intersecting with real-time rendering.
--
Thank you for the kind notes. They were very nice to get. How was your weekend? Mine was not great, but I guess the mentally healthy thing to do is to remember that tomorrow's another chance to try again. Hopefully the week and the coming weekend will be better.
Best,
Dan