Werd I/O

"Your career is not your job. It’s the humans you help along the way."

I feel this in my bones. The people who have helped me in my career have made an outsize difference to my life, and I hope to make the same kind of difference in other peoples'.

The best kind of leadership is servant leadership, which is what I try to embody in my roles: not just to the people who report to me, but the people around me. I want to be helpful. I want to push great people along. And I'm eternally grateful to the people who did that for me. #Business

[Link]

"Until Substack, I was not aware of any major US consumer internet platform that stated it would not remove or even demonetize Nazi accounts. Even in a polarized world, there remains broad agreement that the slaughter of 6 million Jews during the Holocaust was an atrocity. The Nazis did not commit the only atrocity in history, but a platform that declines to remove their supporters is telling you something important about itself."

I'm one of the people who canceled their Platformer membership - not because I don't appreciate Casey Newton's great work, but because I don't want to support a platform that behaves in this way. I'm hopeful that he'll relocate and that I can subscribe again - or that Substack will recant. #Media

[Link]

Charles Stross on tech oligarchs trying to build the torment nexus:

"SF authors such as myself are popular entertainers who work to amuse an audience that is trained on what to expect by previous generations of science-fiction authors. We are not trying to accurately predict possible futures but to earn a living: any foresight is strictly coincidental."

The CIA famously funded the Iowa Writers' Workshop to guide American literature in a non-didactic, less revolutionary direction. I wonder if there's a future in funding science fiction writers to create stories with more utopian themes for leaders to draw inspiration from in the future. #Culture

[Link]

Among other things, this campaign site has a useful list of US retail stores that use facial recognition right now (for example, Home Depot) as well as some that are thinking about it - and some that definitely won't (thank you, Costco).

"Your face should not be scanned, stored, or sold just because you walk into or work at a store. Retailers justify using facial recognition to protect and predict their profits, but this technology puts workers in danger, exacerbates bias, and amasses personal data. Retailers across the country that are exploring this invasive technology should know that prioritizing profit over privacy is wrong."

Co-signed. #Technology

[Link]

I really like this exploration of what it might mean to build a sense of ambient togetherness on the web.

"We currently have no visual, audible, tactile, spatial, or embodied awareness of one another. We also have no awareness of the other people reading this post, even if they're doing it at the exact same moment."

Some of these demos are distracting or not quite right, but they're experiments - and experimentation is exactly what we need. Relatedly, I'm excited about PartyKit as a way to easily build these kinds of experiences.

Maybe I should build something into my own website? #Technology

[Link]

"Of course there will be examples of AI-generated misinformation, bots, and deepfakes during various elections next year. But the key question is how politicians will be using these tools."

This is it: misinformation and disinformation threats are not some nameless force. Sure, there are people out there who will gamify outrage for profit, regardless of truthfulness - but that's always been true. The real harm is conducted by people with power. It's a human, societal problem, not something that can be fixed with technology.

The best fix? Great journalism that speaks truth to power and actually calls politicians out - both on their claims and on their truthless strategies. #Media

[Link]

"Meta’s policies and practices have been silencing voices in support of Palestine and Palestinian human rights on Instagram and Facebook in a wave of heightened censorship of social media amid the hostilities between Israeli forces and Palestinian armed groups that began on October 7, 2023."

This includes posts about human rights abuses. Human Rights Watch found that this censorship was "systemic and global", and often led to removal of content that didn't contravene any content policy.

This isn't new: "In a 2021 report, Human Rights Watch documented Facebook’s censorship of the discussion of rights issues pertaining to Israel and Palestine and warned that Meta was “silencing many people arbitrarily and without explanation, replicating online some of the same power imbalances and rights abuses that we see on the ground.”"

Following that and other reports at the time, Meta promised to address these concerns. It appears that it's fallen far short of doing so. #Technology

[Link]

Some said it would never happen: Mickey Mouse is in the public domain. Or, at least, the very earliest version of him is.

As the BBC points out: "It means creatives like cartoonists can now rework and use the earliest versions of Mickey and Minnie." Disney warns that it'll still protect its copyright on more modern versions, so artists will need to be really creative - but I expect to see some pretty subversive work over the next year. #Media

[Link]

I really like this: a starter pack of blogs to follow if you're new to RSS. Some of them are new to me (and others have familiar authors but seem to have fallen off my list). What better way to start the new year on the internet than subscribing to independent writers again?

One of my projects in the new year is to put together a blogroll - something I'm now convinced every website should have, so that readers can discover new subscriptions organically from people they're already reading.

Perhaps, though, a blogroll is the wrong model, and it should be a regular post like this? That could be fun - Follow Friday for blogs. Hmm.

Anyway, these are great, and I'm grateful to Matt Webb for kicking this off. #Technology

[Link]

There are some jaw-dropping infringements here, including an image where DALL-E apparently copies the entire Pixar universe from the single two-word prompt, "animated toys".

It's impossible to hand-wave this away. Even if you don't think the New York Times case has merit, it's pretty obvious that generative AI can infringe copyright even when you don't ask it to, and without notifying the user. As noted in the references, it's a big ask to then push liability for infringement to the user. It's inherent to the engines.

As the author notes: "My guess is that none of this can easily be fixed." Indeed. #AI

[Link]

"I’d do this even if no one read it. Blogging, for me, is the perfect format. No restrictions when it comes to length or brevity: a post can be a considered and meticulously composed 3,000-word essay, or a spurted splat of speculation or whimsy. No rules about structure or consistency of tone. A blogpost can be half-baked and barely proved: I feel zero responsibility to “do my research” before pontificating. Purely for my own pleasure, I do often go deep. But it’s nearer the truth to say that some posts are outcomes of rambles across the archives of the internet, byproducts of the odd information trawled up and the lateral connections created."

Blogging, to me, epitomizes a lot of the promise of the web. I love it too. And I have no plans to stop.

(An outdated format, though? How dare you!) #Media

[Link]

I like this breakdown of different positions on the open social web: a broad set of things that people who want a big fediverse advocate for, and one for people who want a smaller, safer fediverse.

I'm mostly in the "big fedi" camp. I want the open social web to be as wide and varied as the web itself: a place where any kind of community can erupt and be compatible with all the other communities and still have its own rules and culture. I want supporting fediverse technologies to be as obvious a need as supporting HTML, used by everyone from hobbyists to giant megacorporations.

That doesn't mean that giant megacorporations are my favorite kinds of entities at all. But I think we all gain when open standards are widely supported. A rising tide lifts all groups.

Overall, I guess the answer, for me, is "both". We need the big wide fediverse. But we also need safety and protection, particularly for vulnerable communities. Growth for growth's sake is not a goal; supporting and empowering is. #Technology

[Link]

OpenAI feels a bit like Napster: a proof of concept that shows the power of a particular experience while trampling over the licensing agreements that would have been needed to make the whole thing legal.

The Napster user experience eventually led to our streaming music present: you can draw a line from it directly to Spotify and Apple Music. I expect we'll see the same thing in AI. We know what's possible, a lot of people are excited about it, but it'll take someone else to put the legal agreements in place to actually make it work. (If I had to guess, that company starts with an "A", but it could be a newcomer.)

Once again, the argument that training an LLM is no different to someone reading the same material falls short. Unlike OpenAI, I have to pay for the content I read, and like OpenAI, if I start spewing out large portions of New York Times stories under my byline, I'll end up in court.

I don't know whether OpenAI itself will last. But I am certain we'll see powerful LLMs offered as a service in the future, underpinned by real content licensing agreements for their training data. #AI

[Link]

"Dutch Jews who survived the death camps and returned to the Netherlands were for years monitored by the Dutch secret service because they were considered to be extremists and a danger to democracy." There's no context that can possibly make this acceptable.

The Netherlands is still struggling to come to terms with the way it treated Jews who returned home - and hasn't really reckoned with how enthusiastically many Dutch people supported the incoming regime during the war. It's tremendously sad that it's still trying to make excuses for this behavior today. #Democracy

[Link]

"Your individual risk depends on a large number of factors, but there's a consensus: one out of every ten people who catch Covid go on to develop Long Covid. It's not a one-time risk, either. You face these odds every single time you catch the disease." And that's just the start of it.

Telomere attrition is a side effect that caught my eye, given the telomere dysfunction that runs in my family. The virus ages you - including by compromising your immune system - but it's even worse if you're already immunocompromised or prone to premature ageing.

This piece goes further into the side effects and issues with covid. It's something we need to continue to fight; it's a pandemic that's far from over. Glossing it over in the name of proximal economic stability is short-sighted. People are dying, and even when they're not, the effects can last a lifetime. #Health

[Link]

Technology news tends to center on the global north, but the implications of technology are global. Rest of World does a wonderful, necessary job shining a light on those stories.

What I particularly appreciate about its coverage is that it's not just critical (although there certainly are critical stories - don't skip learning about nickel mining or the economic effects of digital nomads). There are stories of technology-driven empowerment here too, often in surprising ways. #Technology

[Link]

If the Markup is jealous of another newsroom's coverage, you know the stories are good: the Markup has consistently been the most important outlet for investigative technology journalism.

There's a special mention here for 404 Media, which has also been a fantastic addition to the tech ecosystem. I'm grateful that both exist. Both outlets need our support. #Technology

[Link]

I enjoyed Get Back and They Shall Not Grow Old, but what's obvious to me after reading this piece is that these upscaled films will look as outdated in 20 years as CGI from 2003 does today. They'll look like cartoons. Not without value, but nothing close to the intended naturalism.

While I think there's still some value in pieces like those two - anything that makes the past more real so we can learn from it more closely works for me, even if it's not going to be as effective a few decades from now - I'm less excited by upscaled True Lies. Give me the imperfect film grain I remember from my childhood. #Media

[Link]

It took me several attempts to get into Bridge, but finally, this week, I picked it up again and was sucked in. There are plenty of other novels about traveling multiple universes to see the other yous, and Beukes knowingly stops to play with those expectations. The real story here is about loss, and the memory of a person vs the person they really were. But there's a lot of good science fiction fun to be had along the way. #Fiction

[Link]

That an AI model trained on Google Street View photos can look at a picture and figure out where it is isn't much of a surprise, but it's still jarring to see that it's here.

I think the real lesson is that AI undermines security through obscurity, which any security professional will tell you is not a sound approach. It's not enough to assume that information is hidden enough to not be usable; if you want to remain private, you need to actually secure your information.

This has obvious implications for pictures of vulnerable people (children, for example) on social media. But, of course, you can extrapolate: public social media posts could probably be analyzed for identifying details too, regardless of the medium. All of it could be used for identity theft or to cause other harm.

A human probably isn't going to painstakingly go through your posts to figure out information about you. But if it can be done in one click with a software agent, suddenly we're playing a whole other ball game. #AI

[Link]

A sober breakdown of what it may mean for Meta's new social network to finally join the open social web (aka the fediverse).

For many people, this has been a hard pill to swallow: while it's clear that Meta has been a human rights disaster, its embrace of open social web protocols is a vindication and (if you'll pardon the double meaning of the term) a platforming of that movement that may lead to the accelerated growth of the open social web itself.

I would like to see more social networks - both new and established - join the open social web. The biggest thing that worries me is having a single whale in the room that can, in effect, dictate the evolution of the protocols in its favor. A multi-polar social web would be a much more user-centric place (just as the web is at its best when there are multiple major browsers). #Technology

[Link]

I have so much respect for what everyone at The 19th does. It has one of the best work cultures I've ever encountered, and Amanda is a huge part of that. I'm deeply glad to have worked their and to have worked with her.

The things she talks about here are lessons that can and should be learned by newsrooms, but also by organizations across industries.

And I hope they are. Everyone deserves to work in an inclusive, responsible, transparent, empathetic workplace with a strong culture of care. It shouldn't be down to one non-profit newsroom to do this; it should be everywhere. #Media

[Link]

"An investigation by ProPublica and Columbia University’s Tow Center for Digital Journalism shows how false claims based on out-of-context, outdated or manipulated media have proliferated on X during the first month of the Israel-Hamas conflict."

"[...] We also found that the Community Notes system, which has been touted by Musk as a way to improve information accuracy on the platform, hasn’t scaled sufficiently. About 80% of the 2,000 debunked posts we reviewed had no Community Note. Of the 200 debunked claims, more than 80 were never clarified with a note."

So here's the problem. The question is, on a massive online service, what exactly can be done about it. #Technology

[Link]

I think this is right: AI companies, and particularly OpenAI, have a crisis of trust with the public. We simply don't believe a word they say when it comes to privacy and respecting our rights.

It's well-earned. The way LLMs work is through training on vast amounts of scraped data, some of which would ordinarily be commercially licensed. And the stories AI vendors have been peddling about the dangers of an AI future - while great marketing - have hardly endeared them to us. Not to mention the whole Sam Altman board kerfuffle.

I think Simon's conclusion is also right: local models are the way to overcome this, at least in part. Running an AI engine on your own hardware is far more trustworthy than someone else's service. The issues with training data and bias remain, but at least you don't have to worry about whether your interactions with it are being leaked. #AI

[Link]

Interesting study. Although I think there are enormous incentives for people to report this way, I'm sure there's a lot of truth to it.

It sounds like the big issue is the kind of informal ideation and conversation that happens really easily when you're in a room with someone but is rarely raised in a remote environment. It's not that ideas, reflections, and back-and-forth can't happen - it's that there's no really great medium for them to happen on. Slack doesn't capture it; emails and documents are too formal. And a brainstorm is an event rather than a whim.

I don't think people stop thinking creatively when they're at home. But there isn't a great forum for them to raise those creative ideas. That's both frustrating for the person and counter-productive for the organization.

So what does a better medium for this look like? I have some thoughts. #Business

[Link]