View this email in your browser

Dark Matter: 18 July 2018

BOS to DET to STL. Flights are conducive to publishing.

The most-clicked links in last week's Dark Matter:

  1. The design ethics course offered by Mike Monteiro / Mule
  2. Spencer Wright's 2014 project, The Public Radio
  3. Neils Hoven on the limits of analytics in game design

This week in keyword etymology:

This week, (please) begin here: Rodney Brooks has published a four-part series tracing the evolution of AI from early forays to the doorstep of the superintelligence(s) we're staring down today. It's one the best things I've read in 2018 — and absolutely worthy of the considerable time required to read it. No technical knowledge or background required.

* * *

If you missed it, Ben Fry and the team at Fathom released Fakebook : a navigable index of Russia-linked Facebook ads designed to sow dissent and racial animus, with annotations describing the keyword and demographic targeting for each ad instance. It's absolutely unnerving, and a brilliant bit of information design.

* * *

From an O'Reilly blog post on 'good data science' by Mike Loukides, Hilary Mason, and DJ Patil, this:

Moving fast and breaking things is unacceptable if we don’t think about the things we are likely to break. And we need the space to do that thinking: space in project schedules, and space to tell management that a product needs to be rethought.

This seems to be a consistently-forgotten part of the conversation around data and ethics: how we operationalize belief systems. This is epidemic in time-and-materials models.

* * *

Two days ago, Redditor Sinofis posted a link to a project that uses javascript in-browser to build short, shareable deepfakes using a handful of simple mapped celebrity faces. I'm partial to Nic Cage mapped to Donald Trump video clips. Wait, deepfakes?

* * *

This is brilliant: Chinese Whispers is a project by Saurabh Datta that uses a series of Arduino-powered relay, recording, and playback modules to demonstrate the ways in which audio signals degrade as they're repeated by imperfect algorithms. Watch the video: it makes a lot more sense.

* * *

If you've not seen it, spend a few minutes with TensorFlow's browser-based machine learning example library, called Seedbank:

a place to discover interactive machine learning examples which you can run from your browser, no set-up required. Each example is a little seed to inspire you that you can edit, extend, and grow into your own projects and ideas, from data analysis problems to art projects

This week in 'eventual consistency':

From a tweet last week by Lambda CEO (and Y-Combinator alum) Austen Allred:

We're going to try an experiment. Our full iOS + Computer Science course (30 weeks) + a MacBook you can keep + free housing in San Francisco for the full 30 weeks. You pay $0 until you're making $50k+, then 17% of salary for 3 years, capped at $40k. We'll select 10 students.
This isn't an entirely new proposition: Learner's Guild tried something similar, and Sam Ford was excited about a similar, Kentucky-based program when we spoke last year. I've always found the model pretty cynical, frankly.

* * *

From a Colin Nagy interview with W. David Marx, author of 'Ametora: How Japan Saved American Style', published this week in Leanluxe:

The book uses the rise of menswear in Japan as an example of how culture starts from scratch, how brands and media work to push trends, how subcultures reappropriate looks, how globalization works, and how the process of importation/explanation changes the nature of the culture. On that last point, the fact that the Japanese brand VAN Jacket had to bring in Ivy League clothing to a country that had never seen anything like it, meant that they had to explain everything with a bunch of strict rules.

* * *

Spend a few minutes with Anil Dash's not-too-technical explanation of how he programatically unfollowed, well, everyone (and how the Twitter algorithm responded when he did). 

* * *

William Finnegan — author of the brilliant Barbarian Days — has written a terrific profile of NYC Transit Authority head Andy Byford, for The New Yorker. To excerpt liberally:

As Byford travels the city—introducing himself to train crews, bus drivers, tower operators, mechanics—he’s trying to fire people up. Some of his impromptu pep talks seem to delight front-line workers, others not so much. In a lunchroom for drivers at the Mother Clara Hale Bus Depot, in Harlem, he bombed with a sparse crowd of dudes who just wanted to get back to their backgammon. Byford, unfazed, headed into the next room, where a general supervisor of transportation gave him a huge smile. “Where’d you start?” he asked her, a habitual question of his. Bus operator, she said. Twenty-two years ago. “Great!” They bumped knuckles. Byford often gets quizzical looks from employees when he talks about “this company.” The looks seem to say, “What company? I thought I worked for the M.T.A.”

This week in global usability benchmarks:

Adam Greenfield (who I very much wish published more frequently) on the simple joys of the Ember connected coffee mug:

Ember gets some things right, and when it does, they tend to be very right. By far the most important of these is that it works as a mug, prior to the question of any networked or interactive functionality.

* * *

I quite liked Rachel Grossman's piece for Boxes and Arrows on the interpersonal, complex, intangible human dynamics that profoundly shape design outcomes. This, in particular, stood out:

By their very nature, the intangibles are hard to assess. It’s hard to understand where communication breaks down, and why. It’s hard to know whose voices make it into design, and how far; it’s hard to know who is involved in strategy definition and how well a chosen strategy does (or doesn’t) influence outcomes. Because these things are so hard to quantify—and sometimes hard to decipher at all— the ways we communicate, collaborate, and think are slippery and are easily subject to personal bias.

* * *

Here are the reasons that the Government Digital Service doesn't publish PDFs — and why you probably shouldn't, either. Proprietary file formats are friction, friends.

* * *

Finally, I've referenced the SUS (System Usability Scale) here on several occasions — a little technical for a lot of design-focused UX types, but a delightfully simple approach to measuring acceptance and usability of tools and experiences. This week, Jeff Sauro unveiled item-by-item benchmarks for good and great scores on each of the scale's ten criteria. Spend a few minutes with his explanation of the updates.

Until next week.

Copyright © 2018 Dark Matter, All rights reserved.

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.

Email Marketing Powered by Mailchimp