🌀🗞 The FLUX Review, Ep. 105
June 22nd, 2023
Episode 105 — June 22nd, 2023 — Available at read.fluxcollective.org/p/105
Contributors to this issue: Neel Mehta, Boris Smus, Ade Oshineye, Dimitri Glazkov, Erika Rice Scherpelz, Justin Quimby, a.r. Routh
Additional insights from: Gordon Brander, Stefano Mazzocchi, Alex Komoroske, Robinson Eaton, Spencer Pitman, Julka Almquist, Scott Schaffter, Lisie Lillianfeld, 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.
“At times, the solution to a maze is to reduce it to embers and walk straight through the ashes.”
― Mary Doria Russell
🌳💡 Three mazes
Different problems need different solutions. Self-evident, right? However, these differences can be hard to spot when we’re in the midst of the problem.
Imagine problems as mazes. From above, we can tell a lot about a maze: its size, its complexity, if it even has an exit. When we are inside a maze, though, we know a lot less. We know where we are and, if we have a good memory, where we’ve been. But we lack the perspective to know what’s coming.
Some problems that appear to be mazes are actually labyrinths. They are full of twists and turns and can take a while to traverse. But ultimately, we can get through them expeditiously by following the known path. These represent areas where there are plenty of best practices for solving our problem. There’s no need to learn. If our organization is primarily charged with these kinds of problems, we need to emphasize establishing best practices and following the rules.
True mazes have many paths. Yet even when the correct path is unknown to us, we can still validate when we’ve reached the exit. We might not know if it’s the best or only solution, but at least we’ll know it is one. In this sort of maze, we need to have a good algorithm for traversing the maze and a good memory to keep track of our progress.
These mazes are kind learning environments. We can rely on what we’ve seen. We get high-quality, timely feedback. We still need to solve the problem, but we can be confident in our ability to make progress. If we are an organization that is aiming to solve these kinds of problems, we need to accumulate expertise and place cultural emphasis on preserving it.
But some mazes shift. Their layout changes with no obvious rhyme or reason. There might be many exits or none at all. We may exit the maze only to find ourselves somewhere unexpected; not all valid paths lead to the results we seek. No fixed algorithm will help us navigate this dynamic landscape. We must send a multitude of cheap probes to explore. We will never see most of these probes again, but that’s okay. They are just a way to make sense of the maze. With this information we can spot patterns in the shifting layout… and perhaps find a desirable exit.
These mazes are wicked learning environments. We can’t rely on consistency with prior experience. We don’t get reliable feedback. The situation may change even as we try to understand it. In this dynamic environment, we need mental flexibility and a desire to search for insight. Our organization must cherish learning in a way that gives us the room to fail yet always demands that we learn something new from each failure. We must be willing to accept a solution that looks different from our original plan.
Curiously, the best postures to take in each of these environments may be in conflict. The discipline necessary to traverse a labyrinth will irk the experts who want to utilize their skill and experience. The experts will have trouble accepting the failure necessary to explore a dynamic maze. Intrepid explorers will struggle to stay the course of a tried and true solution.
The challenge with mazes is that we can’t always see what type they are when we’re dropped in the middle. We may or may not have been lucky enough to land somewhere with clues — a three-branch intersection, a moving wall, an exit. We need to pause and make sense of our environment so we can choose a strategy for moving.
When an organization is experiencing a cultural upheaval, it might be worth checking to see if there was a shift in the kinds of problems it’s facing. Perhaps the maze we thought we were in is now something else.
🛣️🚩 Signposts
Clues that point to where our changing world might lead us.
🚏🎷 “Only human creators” will be eligible for Grammy Awards
The organization that gives out the Grammy Awards recently updated its rulebook to say that, “a work that contains no human authorship is not eligible [for the award] in any categories.” Humans can still use AI to create songs, but the human must make a “meaningful” contribution to the final product, such as in the lyrics or performance.
🚏👾 Some subreddits went “NSFW” to protest API pricing, then Reddit fired their mods
As the protests over Reddit’s steep new API pricing scheme intensified, some subreddits switched themselves to “NSFW” mode, in large part because Reddit doesn’t run ads on (and therefore can’t monetize) such forums. Reddit has fired back by ousting many of these subreddits’ moderators, saying that “incorrectly marking a community as NSFW is a violation of… our Moderator Code of Conduct.” Some of these volunteer mods have been replaced with new mods — and some communities are quite unhappy with the move.
🚏🍃 Wind and solar generated more power than coal in the US
From January to May of this year, wind and solar together generated more electricity than coal in the US, marking the first time in American history that these renewables have beaten out coal over a five-month span. Coal has been on the decline in the US since 2008, and though the Ukraine war gave it a bit of a reprieve thanks to a spike in natural gas prices, 2023 has seen a returned drop in coal’s usage. (Coal still accounts for 55% of power sector emissions in the US, though it makes up just 20% of total power generation.)
🚏🥵 Electricity prices soared to 70x normal levels amid a Texas heatwave
According to ERCOT, the organization that runs Texas’s energy grid, wholesale power prices in Texas soared as high as $4500 per megawatt-hour (MWh) on Tuesday, This is orders of magnitude higher than the five-year average price of energy in the region, which was just $66 per MWh. The price spike was largely due to a jump in power demand as Texas struggled under a heatwave, with Houston hitting over 101 ºF (38.3 ºC); this is almost the inverse of the time when ERCOT’s electricity prices surged during a 2021 cold snap.
📖⏳ Worth your time
Some especially insightful pieces we’ve read, watched, and listened to recently.
You Can’t Solve AI Security Problems With More AI (Simon Willison) — Argues that no amount of prompt engineering will be able to stop prompt injection attacks, because any clever writer could just tell the LLM to ignore your safeguards. In general, security strategies need to be predictable, rules-based, and reliable — and black-box LLMs aren’t designed for any of those things.
The Dangerous Myth That Human Error Causes Most Car Crashes (The Atlantic) — Argues that the American legal system, insurance companies, and the media usually seek to hold a single person responsible for car crashes, resulting in individuals (whether drivers or pedestrians) shouldering most of the blame. But in doing this, we overlook the far more impactful systemic causes of car crashes: dangerous vehicle and road design.
The Surprising Reason That There Are So Many Thai Restaurants in America (Vice) — Explains the Thai government’s strategy of “gastrodiplomacy”: providing incentives for entrepreneurs to open Thai restaurants across the world (and especially the US) in order to spread interest in Thai culture and tourism. The government even provides pre-fab restaurant plans including menus, pricing, and decor.
Why Did the #TwitterMigration Fail? (Bloonface) — Examines why the influx of Twitter users to Mastodon after Musk’s takeover didn’t really stick. Decentralization hurt Mastodon’s UX and didn’t provide a compelling value proposition to most users; Mastodon instances don’t scale well; and Mastodon’s core community, as the author argues, never wanted Mastodon to be more than a niche corner of the internet to begin with.
📚🌲 Book for your shelf
An evergreen book that will help you dip your toes into systems thinking.
This week, we recommend The Dictionary of Obscure Sorrows by John Koenig (2021, 228 pages).
If Atlas of the Heart laid a foundation for talking about emotion, this book is the advanced course. This is a strange and special book, best consumed non-linearly: pick it up when you’re in a thoughtful mood and want to relish the depth of humanity. What you’ll find is delicate and nuanced writing about being human. Like traveler journals on the spice roads, these detailed accounts of far-flung corners of the emotional landscape enrich our internal menagerie and vocabulary.
The Dictionary of Obscure Sorrows is presented as a dictionary of newly-invented words, perhaps the most famous of which is sonder (“the realization that each random passerby is living a life as vivid and complex as your own”). These words describe clearly felt, but often hard to verbalize, experiences in our lives. These feelings may be fleeting or persistent. They could be a visceral hit in the gut or evoke a feeling of wonder. These words might even help you recognize a feeling that you’ve observed in others without fully understanding.
And, perhaps, some passages will bring tears to your eyes — the cathartic, healing kind of tears that fall when we come in contact with boundless kindness.
🔮📬 Postcard from the future
A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.
// How might AI change the way companies form?
// An in-person office hours session of the Fall 2025 Z Combinator program. The founders of a dozen startups sit and listen with rapt faces as one of the partners begins to talk.
“Hi everyone! Welcome to the penultimate weekly office hours. As you know, we selected you to join our elite program because we believe you have the skills and potential to build wildly successful companies. We have spent the past several months providing you with tools, wisdom, guidance, and feedback to help refine, pivot, and redefine your startups.
“I know you’re excited for next week’s session, where we cover our final pitch and unveiling event. However, this week, we’re going to talk about the risks of becoming blind. You’ve made it this far because you believe in your company, your co-founders, and your idea. You’ve made it through periods of doubt and failure that would crush most people’s desire to start their own company. But not you. You are here. You have confidence.
“Group confidence is critical, but it can lead to group-think — heck, even group-faith. This can lead to the enshrinement of certain ideas that ‘shall not be questioned.’ And if you have those shrines, you have blindspots which competitors can exploit.
“The Catholic Church used to have an official position named advocatus diaboli: the devil’s advocate. This role was one who argued against the canonization, or sainthood, of a candidate in order to uncover any character flaws or misrepresentation of the evidence favoring canonization.
“Every organization needs someone to take opposing viewpoints, but in a small company, that can be seen as disruptive or not buying into the company’s mission. This is particularly unfortunate for small companies like startups because they are often places where if the question is asked, they are nimble and resilient enough to make the necessary changes to illuminate and eliminate blind spots.
“So that’s why we are excited to announce a paradigm shift in how Z Combinator operates: your graduation from Z Combinator is contingent on you implementing some form of ‘Devil's AI-vocate.’ By training an AI on your startup’s Slack/Discord/email messages, code submissions, and any other text related to your startup, you will create a magic mirror that lets you understand all the blind spots no one wants to or can even imagine to point out.
“You have until next week to get it done. Now, who wants to ask the first question?”
© 2023 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.
https://thecynefin.co/about-us/about-cynefin-framework/