🌀🗞 The FLUX Review, Ep. 92
March 23rd, 2023
Episode 92 — March 16th, 2023 — Available at read.fluxcollective.org/p/92
Contributors to this issue: Dimitri Glazkov, Neel Mehta, Lisie Lillianfeld, Erika Rice Scherpelz, Justin Quimby, Boris Smus
Additional insights from: Ade Oshineye, Gordon Brander, a.r. Routh, Stefano Mazzocchi, Ben Mathes, Alex Komoroske, Robinson Eaton, Spencer Pitman, Julka Almquist, Scott Schaffter, Samuel Arbesman, Dart Lindsley, Jon Lebensold
We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.
“The original purpose of a hierarchy is always to help its originating subsystems do their jobs better. This is something, unfortunately, that both the higher and the lower levels of a greatly articulated hierarchy easily can forget. Therefore, many systems are not meeting our goals because of malfunctioning hierarchies.”
— Donella Meadows
🦜🪣 Information sloshing
You’re working on a big collaboration, trying to define and move toward a shared goal. You sense some misalignment and ambiguity, so you schedule some 1:1s. Other pairs of folks also have 1:1s… uh oh! This is trending toward n2 meetings. Next you try group meetings. They feel a little aimless — lots of status updates. You’re learning what everyone is doing, but things still don’t seem to be moving. You still don’t know enough to come up with The Shared Vision. The problem you are experiencing is “information sloshing” — information is moving around, but not really going anywhere.
A common reaction to information sloshing is to try to run the meetings better: clearer agenda, more detailed notes, a new process, less process, more attendees, fewer attendees. But these changes just slosh more information. We’re still not moving forward.
Instead, try shifting your perspective. Rather than trying to solve for the big and amorphous shared vision, pick a narrower slice of the problem and try “information funneling.”
For “information funneling,” start a doc around a particular sub-question and ask everyone to comment with their thoughts on that specific issue (funneling the information relevant to this narrow topic into one place). For each separate sub-question, start a new doc. This organizes conversations around smaller, better-defined issues. These smaller conversations often involve a subset of the stakeholders and can get resolved more quickly. Funnel docs also benefit your future self when you need to onboard new people, remember why the group made a certain decision, or demonstrate progress to your manager.
How do funnel docs move us toward defining a shared vision? Finding common ground, even on just a few points, can narrow the space of what is unknown, clarify constraints, and reveal what other decisions need to be made. From there, the path to defining and aligning on a vision is clearer and more tractable.
Ideally, we can develop a shared vision relatively early to make sure that smaller decisions are moving the project in a productive direction. But when that's not happening and you feel information sloshing, try funneling information into some smaller docs and see if you can get traction from there.
🛣️🚩 Signposts
Clues that point to where our changing world might lead us.
🚏🖲️ ChatGPT helped researchers win $123,000 at a security hackathon
At the “Pwn2Own” hackathon, participants were challenged to break into Internet of Things (IoT) devices. One team of cybersecurity researchers identified several bugs in their chosen target and then asked ChatGPT to write code that’d chain those exploits together. The combined piece of software succeeded all 10 times it was tested, earning the team a cool $123,000 in prize money.
🚏🛒 IKEA is using autonomous drones to count stock in warehouses
IKEA’s parent company has announced that it’s deployed a fleet of 100 blue-and-yellow autonomous drones to fly around its stores after hours and track inventory levels. The drones capture images, videos, and 3D depth scans of each pallet in the store, and then alert employees if items are misplaced or stock is running low. IKEA has previously introduced automated racking systems that have “eliminated the majority of forklifts” in stores, and even an automated warehouse where robots retrieve items that customers ordered online.
🚏📜 New scanning tech can “remove” stains from old documents
In a new paper, a team of computer vision researchers laid out a new approach for digitally restoring old documents that have been damaged by the ravages of time. This technique scans a document pixel-by-pixel to identify different layers, such as the paper, ink, and stamps. It can then identify and delete stains, spots, and ink bleed-through, and it can correct for inconsistent lighting. The researchers say this method promises to make old documents readable for historians without touching (and potentially damaging) the original artifact.
🚏🇨🇳 Beijing’s population fell for the first time since 2003
Beijing’s population fell from 21.88 million in 2021 to 21.84 million in 2022; it’s a tiny drop, but it’s also the first population decrease the megacity has seen since 2003, when it was hit by a SARS outbreak. A major reason is that the number of internal migrants into Beijing (mostly rural villagers looking for work) decreased from 2021 to ‘22. This series of declines coincides with China’s overall population decline; China’s national population famously fell last year for the first time since 1961.
🚏🍰 3D printers can now assemble and bake cakes
Researchers have designed a robot that can create a cake with seven ingredients, more than any 3D-printed food to date. The 3D printer squeezes out jelly, banana puree, peanut butter, and Nutella inside a frame of graham cracker crust. After it “bakes” the cake by blasting it with a laser, the robot adds frosting and cherry glaze on top. In addition to automating repetitive kitchen work, this robo-cooking approach could also build foods with hyper-precise nutritional profiles.
📖⏳ Worth your time
Some especially insightful pieces we’ve read, watched, and listened to recently.
High Variance Management (Sbensu) — Stage actors need to be good at every performance, but film actors have the luxury of multiple takes. Analogously, rather than insisting on perfection every time, managers can embrace an evolution-inspired approach called “creative selection,” encouraging multiple teams to tackle the same problem, prototyping and demoing to one another, and then having a decision maker pick the best approach.
To Teach Computers Math, Researchers Merge AI Approaches (Quanta Magazine) — Describes how AI researchers are trying to teach large language models to understand and solve math problems. Much of it boils down to “autoformalization”: turning mathematical statements in natural language into logical building blocks. It’s analogous to converting human language into computer code.
Who Is Still Inside the Metaverse? (New York Magazine) — Catalogs one writer’s recent adventures in Meta’s Horizon Worlds, reflecting on the kinds of people he met there and what drew them to the VR world. The metaverse is at once lonely (“nobody has any connection to anyone else beyond owning a headset, a weak tie if ever there was one”) and a reprieve from loneliness (“I can’t help noticing how many of the stories tonight are about being alone”).
Introducing Not-Knowing (Vaughn Tan) — Builds upon the concept of Knightian Uncertainty, which differentiates “known risk” from “unknown risk.” Tan suggests we drop the confusing term of “risk,” and instead focus on various kinds of “not-knowing.”
The Planet Closer to the Sun Than Mercury | Vulcan (Astrum) — Explores how 19th-century astronomers became convinced that there was a never-before-seen planet inside Mercury’s orbit; despite the “planet” never actually being sighted, astronomers kept tweaking their models to make it seem possible. The story is a good lesson on confirmation bias, showing that even famed scientists aren’t immune to it.
🔍📆 Lens of the week
Introducing new ways to see the world and new tools to add to your mental arsenal.
This week’s lens: the sloping floor.
After a failed project, an engineering team often conducts a thorough retrospective. They identify the problems that led to the unsatisfactory result, create a plan to solve them, and commit to following it. Nearly as often, though, time passes and the same problems keep cropping up. Despite their best efforts, the team recreates the outcome they had sworn to avoid.
A good metaphor for such cases is a sloping floor. If we place a ball on an even floor, it will stay where we put it. However, should the floor be sloped, even slightly, the ball will move down the slope. On a sloping floor, keeping the ball in place requires effort. If we are not aware that the floor is sloped, we will be constantly bewildered to find it in one particular corner of the room.
A similar dynamic plays out in teams. If we aren’t aware of the particular slope of our floor, we will not know where to put our energy to counteract the gravity that keeps rolling us into a particular corner. The slope can hide in various places, from the team’s cultural quirks to processes and practices to choices of infrastructure.
It is hard to spot a small slope from a freeze-frame snapshot. We might have more success if we watch for the effects of the sloping floor show up in a team’s activities. What things are exceedingly easy for us (downslope)? What things are more difficult (upslope)? If we look for these things, we can start to see the gradient of our world.
© 2023 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.