Weeknotes 2024/26

The big three topics this week were: post-shame society, regulation of technology, and information vetting becomes more important than ever. That Matt Levine quote and someone angrily questions the AI hype by an industry that still fails on much more mundane stuff.

Last week was mostly spent developing content and copy, refining and clarifying, rejecting and rewriting. I still have a lot to learn when it comes to writing content.

So without further ado, let’s get into the link topics.

Post-Shame Society

Finally I know what it is called. I will definitely keep this on the radar as this might also be behind ever-expanding organisational dysfunctions. It’s not just the rich and powerful in the classic sense. It’s the richer and the more powerful in general. It’s not absolute, it’s relative. You see this behaviour spreading among people, who would not be considered rich and powerful. They are just richer and more powerful in a certain cohort, e.g. a company or a business relation. Every time someone makes you think, they have to channel their “inner Musk”, it’s likely they operate in post-shame waters.

It became fashionable to show a lack of transparency and accountability. I sometimes wondered how people aren’t ashamed of themselves, when showing this conduct. Now I know.

As usual, when learn something new, you start to notice it in many places.

💰Trump, Musk, Supreme Court justices exploit post-shame America

The article mostly focuses on the richest and most powerful, but, again, I would argue this has already spread much further and wider.

Fast Crimes at Lambda School

Detailed report on what was happening at Lambda School. It includes quotes from the founder that makes you nod and say “ah, post-shame”.

These Political Nonprofits Spend 90% of Donations on Fundraising — ProPublica

Yet another example of “post-shame”.

Not sure what to make of it yet. It adds to the risks our society is already facing. We can only counter with better conduct and actions.

Regulation of Business and Technology

Last week both the EU DMA and EU chat control law made headlines, with chat control being dismissed (for now). Now they are quite different beasts, but they share the signals showing that regulation has become ineffective.

It’s easy to dismiss regulation due to the incompetence shown by law-makers throughout the process, but I want to stress that regulation or laws only get on the agenda when someone feels that a company or an individual isn’t using their freedoms responsibly.

As such whenever you’re annoyed by the conduct of regulator, EU or otherwise, always remember that the industry could act a little more responsibly and proactively. Why does it always take an external motivation to take action? A little more transparency and accountability would also move the needle significantly.

Chat Control / OSA

Chat control and OSA both are regulations/laws that should prevent spreading CSAM. So at least we have good intentions. The problem is that these regulations deal with symptoms rather than root causes and in doing so open up avenues for multiple other problematic conducts, mostly related to privacy.

Furthermore, they show the lack of understanding of the domain by the law-makers to almost hilarious extents.

Law-makers should focus on the root causes of child abuse to have a significant impact. To put it differently: the creation of CSAM should be the focus, not the spreading of CSAM, because the harm has already been done.

Client-Side-Scanning: Chat Control is Pure Surveillance State

Netzpolitik’s take on Chat Control.

‘Encryption is deeply threatening to power’: Meredith Whittaker of messaging app Signal | Chat and messaging apps | The Guardian

Meredith Whittaker interviewed by the Guardian. In this case concerning Britain’s Online Safety Act and some wider aspects of such laws.

EU Council has withdrawn the vote on Chat Control

EU Chat Control is on hold.

EU cancels vote on child sexual abuse law amid encryption concerns – POLITICO

Politico’s take.

The EU wants to end encryption. It doesn’t end today

Sebastiaan de With’s take, although he has some quotes from EU politicians that make you go “post-shame” again.

Leak: EU interior ministers want to exempt themselves from chat control bulk scanning of private messages - EU Reporter

Supporting the message that some EU ministers lied about the privacy implications evidenced by the exemptions for chat control.

DMA

This is a mostly annoying example of regulation, especially when it comes to Apple. First, their own image building holds them to higher standards of conduct than what they are displaying. Second, many of the elements in the DMA have been discussed in the community for a while. A more cooperative, transparent and proactive approach by Apple could have lead to much better outcomes for their customers in the EU. They dropped the ball on customer satisfaction, that’s for sure. They have no problem caving to much more problematic regulations when it comes to China. On many topics of the DMA they behave like a petulant child.

Apple Intelligence Features Not Coming to European Union at Launch Due to DMA - MacRumors

Some announce features are not coming to the EU due to the DMA. It sounds more like a delay, but you never know. Apple may be trying to change the narrative a bit by showing EU law-makers the unintended consequences the DMA can have.

Personally, I can understand the AI features, but the exclusion of iPhone Mirroring did offend me.

Daring Fireball: Financial Times Reports EC to Charge Apple With Non-Compliance Under DMA for CTF

European Commission is offended by the Core Technology Fee. The world’s most stupid boxing match continues.

Daring Fireball: The EU Is Reaping What It Sows With the DMA: Uncertainty

John Gruber is right on the DMA, but Apple isn’t as innocent as they want to picture themselves.

Daring Fireball: Apple Disables WebKit’s JIT in Lockdown Mode, Offering a Hint Why BrowserEngineKit Is Complex and Restricted

Another signal that regulation is not the best way to handle things.

Information Vetting Becoming More Important

Both when it comes to input and output information. We covered a few links last time concerning the vetting of input information for AI. This is the only way to keep the output of such systems in a high enough quality to be useful. Even then, we need to hone our information vetting and confabulation recognition skills in order to keep decision making sound.

How to Vet Information Before Making a Decision

You could argue this has been exaggerated by AI. See below.

Researchers describe how to tell if ChatGPT is confabulating | Ars Technica

Article on the confabulation recognition skills, aka. output crap filter. Also increasingly important are communication skills to deal with people who do not have this skill and don’t bother getting it, but simple take confabulation as truths (because it’s easier).

Enterprise AI Requires a Lean, Mean Data Machine - The New Stack

A further plea to improve the data game at (large) companies, because crap in, crap out. Make it an organisation-wide responsibility or even priority.

Daring Fireball: Popular AI Chatbots -- Including ChatGPT, Mistral, and Meta AI -- Spread Russian Propaganda (Because of Course They Do)

Guess what went into the training of these?

💬 That Matt Levine Quote

OpenAI was founded to build artificial general intelligence safely, free of outside commercial pressures. And now every once in a while it shoots out a new AI firm whose mission is to build artificial general intelligence safely, free of the commercial pressures at OpenAI.

You may wonder what has changed between earlier splits and the current one? Time. My hunch is that they were dazzled by the charismatic version of Sam Altman and/or others and just bought that safe AI really was the priority. The conduct that lead to the first wave of quitters likely remained and added up to the point where the current wave of quitters couldn’t ignore it anymore. At this point, unless we see some signal of fundamental change at OpenAI, I would would bet that this isn’t the last wave of people quitting due to OpenAI’s safety practices (or lack thereof).

Daring Fireball: Matt Levine on OpenAI's True Purpose

John Gruber quoting Matt Levine.

A quote from Matt Levine

Simon Willison quoting the same bit.

OpenAI’s former chief scientist is starting a new AI company - The Verge

Coverage of yet another spawn by former OpenAI employees, this time Ilya Sutskever.

Safe Superintelligence Inc.

Ilya Sutskever’s new company mentioned in the article.

Everything Else

🤬 I Will 🤬 Piledrive You If You Mention AI Again — Ludicity

Savage, but fair take on the AI hype, with a bit too much foul language for my liking, which detracts from those kernel of truths in there. The quote highlighted by Simon Willison though, is dead on the core problem of literally every company’s AI initiative:

“[…] And then some[one] [expletive] created ChatGPT, and now look at us. Look at us […], spending half of the planet’s engineering efforts to add chatbot support to every application under the sun when half of the industry hasn’t worked out how to test database backups regularly.”

🤖 Exclusive: Multiple AI companies bypassing web standard to scrape publisher sites, licensing firm says | Reuters

Well, who knew I could link to my AI crawlers post almost on a weekly basis now.

🤔 Research: Using AI at Work Makes Us Lonelier and Less Healthy

The insomnia part is curious. From the article I can only assume this is a secondary effect to loneliness and alcoholism.

🥇 JavaScript Framework Maintainers on Unification Potential - The New Stack

Unification should not happen. Say what you want about the JavaScript fragmentation (queue the “a framework per week” joke), the most recent developments work in its favour. While a few years ago it seemed like framework decision subscribed you to a fundamentally different way of thinking about web apps, today there is way more convergence in the mental models behind a specific developer experience. With maybe 2 or 3 frameworks that push the envelope, challenge the status quo and have an impact on the more established frameworks (looking at Solid.js for example).

🤔 Calls to Ban Open Source are Misguided and Dangerous - The New Stack

As if Open Source software as a concept didn’t have enough challenges these days, mostly driven by ignorance, it now gets caught in the wake of Open Source AI discussions, which is a different thing, but gets conflated anyway. Especially when someone calls for a ban without context.

🤷‍♂️ Why does SQLite (in production) have such a bad rep? - blag

SQLite has had a reputation problem for ages and I never understood why. It’s one of the best maintained pieces of software there has ever been and excels in crazy number of use cases. In fact, I can recall one conversation with a manager who dismissed it on the grounds of the name and didn’t even approve using it for an export format instead of a complex and brittle XML mess. It was the point when I realised that regulated software environments might even be dysfunctional because of the regulation.

📚 What is a Personal User Manual? - Future Forum

I have seen critical takes on the personal user manual, esp. in the form of a manager’s user manual, but I still advocate for it. The problem was that in some settings it was weaponised under the pretend of psychological safety to mean “this is how I work and every other way will be seen as a personal attack and bullying and will be reported to HR”. The personal user manual’s biggest pro is the written form that it’s a form of self reflection. In its best form, it can open up a discourse of what healthy collaboration can look like and what’s acceptable.

🙅‍♂️ Why Innovation Heroes are a Sign of a Dysfunctional Organization

I think I read this before, but it’s worth the read. Why? The same mechanics that require innovation heroes will also kill any sort of continuous improvement culture in the organisation. Those are not the same and I have argued for years that removing impediments have more impact than your once a year innovation or transformation.

🖱️ Why employers get "mouse jigglers" that fake working

Great article showing that it takes two to tango. Those “mouse jigglers” are just as much a failure of management and leadership.

🦇 Spatial Computing Hack | Ryan Pickren

Finally a bug that makes you laugh instead of sigh. Warning, if you’re afraid of bats, this is not a laughing matter.

🤓 The Blank Sheet Method: From Passive Reading to Active Learning

As an active reader for years, I have seen variations of this in the past and can attest that it changes the way you interact with a book.

🆔 UUIDv7 in 33 languages

I will likely switch the UUIDs of this site to v7. It was one of my first design decisions that I wanted to use an ID system since I don’t like URL slugs. The advantage of slugs for SEO is debatable (using the URL as a signal is a hack worth nothing) and since one rarely types in a full URL these days, the user experience argument is moot (and was always debatable anyway). I should probably write on the IDs I found on my research. I’m not sure, why I haven’t chosen v7 from the get go, but there is no harm in switching. The old UUIDs stay as they are

🦖 jwz: Mozilla's Original Sin

Mozilla misstepped in so many ways, it’s heart-breaking. Mozilla’s decision to include DRM probably contributed to change in culture that allowed this to happen. His conclusion what Mozilla should be is sound, but I’d add they could be the reference implementation for so much more around internet privacy and standards. And they need to, because search is in serious trouble, which will likely have an impact on their Google search deal.

😅 React 19 Change Angers Some Devs; Vector Database Use Jumps - The New Stack

React 19 made an oopsie. Sometimes you’re stuck between a rock and a hard place. The culprit has to do with <suspense>. The improvement for one use case where it performs pretty poorly, has introduced significant performance regression for other use cases. I’m not too deep into it, but from what I read, they have 2 options: optimise for one case by default, allowing devs to switch to the other behaviour, or rethink suspense all together.