Ben Werdmuller

I’m soliciting prompts for discussion. This piece is a part of that series.

Erlend Sogge Heggen asks:

There’s legitimate reason to be worried about Meta’s P92 project being part of a EEE play against the fediverse.

How might the fediverse community counteract this, perhaps with its own EEE strategy?

We know Meta will attempt an EEE play, but what if we play the reverse UNO card and EE(E) them instead?

Embrace: Carefully federate in a minimum-viable fashion that doesn’t overrun the existing #fediverse.

Extend: Make #NomadicIdentity a reality, so accounts can be moved effortlessly.

Extinguish: In case of misconduct, defederate and provide mass-migration assistance.

First, some quick definitions!

P92 is the codename for Meta / Facebook’s new app that will support the same ActivityPub protocol as Mastodon and its cousins. Users will be able to log in with their Instagram credentials, and one can potentially (but not definitely) imagine it being folded into the mainline Instagram app.

Embrace, Extend, Extinguish was a phrase coined internally inside Microsoft to describe its strategy with respect to the web. The idea was that the company would embrace open protocols, extend them with its own proprietary extensions, and then use its control over those extensions to extinguish competition. In particular, its plan was to do this with HTML in order to cement Internet Explorer as the web browser.

Finally, the fediverse, of course, is the community of small, independently-owned, largely non-profit social networks that interoperate using shared protocols, on which Mastodon is the largest platform.

There is legitimate concern that a company like Meta might attempt to control the fediverse. This is particularly true if they are allowed to create a uni-polar world: one where Meta is the only large company embracing these standards. In that world, Meta can throw hundreds of millions of users at the protocol, and it will instantly become its largest user.

I think it’s helpful to look at how Microsoft’s EEE strategy failed. There were arguably two main factors: antitrust risk and competition.

The Department of Justice sued Microsoft for monopolistic business practices, ultimately leading to a settlement where Microsoft capitulated to changing some of its approach in return for the DOJ dropping its desire to break up the company. It’s not clear to me that this kind of case would or could take place with respect to Meta extinguishing the fediverse; while I’m not a lawyer, I think the argument would probably be that many other social networks are available.

The other thing that hurt Microsoft’s dominance was Firefox. It was a good browser backed by a good community, but that wasn’t the deciding factor; Firefox gained market share because Google pushed it at every possible opportunity. Because Internet Explorer’s dominance was a business risk to Google, and because Firefox was built by a non-profit that was non-competitive with Google’s business, it made financial sense to try and break Microsoft’s stranglehold. Mozilla’s model was stronger than its predecessor Netscape’s had been: whereas Netscape needed to sell licenses, Mozilla’s deal with Google meant it made money every time someone used Firefox to search for something on the web. There was almost no friction to its growth strategy.

This activity led to a resurgence in a healthy ecosystem of standards-based web browsers for years — until Google decided to re-use the technique it had used on Firefox to push its own web browser. Even then, Chrome is a far better standards player than Internet Explorer ever was.

There won’t be hard evidence that Meta is adopting ActivityPub until we see its app in the wild. But if it is, that likely means that it sees the protocol as at least worth experimenting with, and maybe even as a potential threat. That’s a sign of great progress, and everyone involved in building the fediverse should feel good about it.

If Meta wants to own the fediverse, this isn’t a battle that will be primarily won with features or technology. Easy-to-use platforms, nomadic identity that easily lets you move your presence from one provider to another, and assistance will all be essential, but they’ll be table stakes. (If Meta is working on the platform today, it’s probably also too late for truly nomadic identity to make a difference.) To really stand a chance, the fediverse will need the kind of marketing and go-to-market support that Firefox enjoyed back in the day. Which may mean support from another large player that considers Meta’s ownership of the standard to be an existential risk.

It’s hard to see who that might be. Twitter is now the incompetence wing of the incompetence party. It’s highly unlikely that networks like Pinterest care. Microsoft’s platforms are tightly bound to its ecosystem, with access control at their core; I don’t see LinkedIn joining the fediverse any time soon. Google has fallen on its face every time it’s tried to build a social network, and runs YouTube as a separate entity that strongly benefits from closed ads. Salesforce might consider it a risk, as it provides social tools for businesses, which are easier to build and sell on an open social networking standard. Some of these entities might consider the fediverse to be worth exploring — but there’s no clear technology backer. Cloudflare actually did provide its own Mastodon-compatible platform that runs on its CDN, but it hasn’t seen anything like wide use. Medium has embraced Mastodon but has not deeply built support into its existing platform.

Perhaps media companies, who generally live and die on the size of their audiences, and have often been beholden to the large social networks, might find themselves interested in embracing a social networking federation where they have more say and control. The rise of the fediverse certainly is a de-risking of their business models. But I don’t think they see it yet; nor do I think they consider it their place to pick a winner. (Nor should it be, really, in practice.)

Perhaps there can be another kind of backer: an entity that sees the existential thread centralized control of social media poses to democracy itself. We’ve already seen how, left unchecked, centralized companies like Facebook incite genocides and throw elections. The fediverse can be an antidote to these trends — if we see it as a set of collaborating communities rather than simply the technology alone. The erosion of democracy, like monopolistic abuse of power, are human problems with human solutions rather than technological ones. Foundations and philanthropists may choose to provide this level of support, if they continue to see Meta as a threat to democracy.

Building features will not protect the fediverse from being extinguished, although they may provide a useful baseline. It’s going to take a whole different level of strategy, relationship-building, deal-making, and movement-building. I believe the fediverse is capable of doing this, as long as it doesn’t mistake building software for making true progress.

Via Jeremy Keith, Future-First Design Thinking:

Incorporating non-human personas into the design thinking process would allow us to embody the essential elements that constitute our environment, such as air, trees, water, and land. These personas can serve as tangible reminders of the interconnectedness between our design decisions and the health of our environment, a living entity that deserves our protection and consideration.

I’m sorry to say that I don’t trust it.

My bias: personas are harmful. A persona is an amalgam — a fictional person, really — that is supposed to be drawn from extensive user research. The problem is, the process of drawing up a persona always requires a degree of subjective invention, regardless of the amount of research that went into it. It is always a reflection of the biases on the team.

In contrast, POV statements that describe a single person who has been interviewed, in combination with direct transcripts and photographs, allow you to use a non-invented example to build your product — and, at least theoretically, go back and use that person as your referee for product decisions as you continue to build. While condensing a person’s interviews into a POV does require some invention, it isn’t a statistical exercise. You can always go back and ask.

It’s very difficult to do this with a tree. By its nature, then, all you can do is invent — and potentially excuse all kinds of activity because your fictional tree persona approves (or, more likely, has blind spots).

All of this said, I do think it’s vital to include future-facing sustainability in design thinking frameworks. (Design thinking is prone to colonial thinking, so considering distributed equity is vital too.) My proposal was to add sustainability to the desirability, viability, feasibility lenses; it’s not necessarily right, but it’s something to consider.

I’m soliciting prompts for discussion. This piece is a part of that series.

An anonymous reader asks:

‌Do the good guys need to stoop to the tactics of the bad guys in order to avoid losing, or is there still value in holding oneself to certain standards?

First: who are the good guys and the bad guys? Can you be sure that’s true? One of my all-time favorite comedy sketches is Mitchell and Webb’s “are we the baddies?” skit. (The writers room for That Michell and Webb Look included Succession creator Jesse Armstrong and Oscar winner Olivia Colman. It’s well worth checking out if you’re new to it.)

I don’t think it’s worth getting into a discussion about objective good or objective bad, or the shades of morality in between. I’m not a moral philosopher and will not do that topic justice. But I can tell you who I think needs to be supported and who needs to be hindered in the current moment in order to create the world I’d like to see. Hopefully that’s a good enough analog.

My values are that equality is a virtue, and that everybody should be able to live a good life, regardless of their background or context. Fundamentally, nobody is more valuable than anybody else. That means I’m in favor of social programs like welfare and universal healthcare; I’m pro-choice; I’m heavily against the death penalty; I believe strongly that there are serious structural inequities in society that must be redressed; I value cosmopolitan spaces, immigration, multiculturalism, and inclusion. I think an abundance of guns and a militaristic foreign policy are tools of dominance and intimidation that must be relegated to the past. I was and am in favor of covid lockdowns and restrictions.

You might not agree with me on some of these things. That’s completely fine! One of the core tenets of democracy is that we can have opposing views and debate them — and that there’s open access to the reporting and data we need to make informed decisions.

From this, you can probably discern the ideas that I think are harmful or “bad”. Monoculturalism; impeded debate; social hierarchy / structural inequity; fascism; violence; anti-intellectualism; nationalism; dominance. The people who would ban books from libraries, eradicate open journalism, impose their worldview and ambitions through force.

Some of the tactics used by proponents of these ideas are abhorrent to the point where “stooping to their level” would make a movement indistinguishable from them. Voter suppression is one; storming the Capitol is another; book banning a third; tricking asylum seekers into flying to Martha’s Vineyard as a political stunt is obviously a fourth.

Those values are actually in the statistical minority: in America, in Britain, around the world. Undemocratic tactics are required to keep those people in power.

But here’s one tactic that I do think we can do better at. Fascists have always been great marketers, and explicitly used marketing tactics; Edward Bernays influenced the propaganda Goebbels produced for the Nazis, for example. “Make America Great Again,” though recycled from Reagan, complete with its recognizable Nazi-red hats, has proven itself to be an enduring brand (albeit one that turns my stomach). More importantly, they’ve made promises about what they will give to their base.

Progressives have not done as great a job at marketing themselves - in any arena. Lately, we haven’t so much painted a picture of how the world could be as pointed at the bad stuff and said, “at least we’re not that.” I believe Joe Biden was mostly elected by not being Donald Trump. New social networks gain popularity by not being Twitter. At-least-we’re-not-that isn’t actually a terrible tactic right now, given how bad that really is, but it can’t last forever, and it’s quite a long way from hope and change.

There needs to be a simple message about how regular peoples’ lives will improve in a progressive world. Bill Clinton, who was not progressive by most measures, was right when he said “it’s the economy, stupid” — the message isn’t about how the world will be fairer, but for each voter must be about how the world will be better for them. That story needs to be told, simply, but with all the skill that anyone can muster. Because it really is a better world.

Thinking one’s values are ideological better does not absolve a movement from having to address, directly, how it will make the world better for every person. When that work is really done, it will be a stronger message. For one thing, hopefully, unlike the opposition’s, it will actually be true.

Inspired by Manton’s list of countries and states he’s been to, I thought I’d come up with my own.

I’ve lived in five:

• The Netherlands 🇳🇱
• England 🏴󠁧󠁢󠁥󠁮󠁧󠁿
• Austria 🇦🇹
• Scotland 🏴󠁧󠁢󠁳󠁣󠁴󠁿
• The United States 🇺🇸

In addition to those, I’ve been to (excluding those “I’ve technically been to here but I really haven’t” airport layover visits):

• Albania 🇦🇱
• Belgium 🇧🇪
• Canada 🇨🇦
• Denmark 🇩🇰
• France 🇫🇷
• Germany 🇩🇪
• Greece 🇬🇷
• Italy 🇮🇹
• Liechtenstein 🇱🇮
• Luxembourg 🇱🇺
• Mexico 🇲🇽
• Russia 🇷🇺
• Spain 🇪🇸
• Switzerland 🇨🇭
• Vatican City 🇻🇦
• Wales 🏴󠁧󠁢󠁷󠁬󠁳󠁿
• Yugoslavia, when it was Yugoslavia 🇧🇦🇭🇷🇲🇰🇲🇪🇷🇸🇸🇮

There are some serious and obvious omissions on this list. I’ve never been outside of Europe and North America! And even in Europe, I’ve never been to Iceland, Ireland, Sweden, Norway … and so on. There are so many places on my bucket list — not least Indonesia, where so many of my family comes from. One day, I would also like to explore my roots in Ukraine; I hope it’ll be safe to do so.

I’ve driven across the US three times, so I’ve been to markedly more US states and the District of Columbia. In fact, every state aside from Kansas, Kentucky, Missouri, and South Dakota. I’ll have to correct those soon.

I’m soliciting prompts for discussion. This piece is a part of that series.

Michael Kriegh asks:

‌Do you think AI is an evolutionary trend for intelligence in the universe? If so, what do you imagine that trend will look like in 50 years? 100 years? If not, why not?

There’s a body of work surrounding the potential of truly artificial / alternative intelligence on human development. You can read about some of that on Michael’s site. However, I’m going to take another, simpler approach to answering this question.

Most, if not all, of the software we call AI is not intelligence at all. They can’t think, or reason, or discern. They’re pattern-matchers. Arthur C Clarke’s third law states that any sufficiently advanced technology is indistinguishable from magic; any set of sufficiently advanced heuristics is indistinguishable from intelligence. Just as technology is not magic, software is not intelligence. It’s a tool.

Still, this doesn’t mean it can’t have an evolutionary impact on intelligence. It’s worth considering how new tools have affected the evolution of intelligence in the past.

For example, stone tools:

The development of sophisticated stone tools, including sturdy cutting and sawing edges, is considered a key moment in human evolution, as it set the stage for better nutrition and advanced social behaviours, such as the division of labour and group hunting.

These behaviors, according to the cited study, evolved alongside language, as both required more complex thought. Language enabled coordination; tools enabled nutrition and the development of better and better equipment that eventually allowed humans to travel around the globe. In turn, we adapted for the new environments we found ourselves in.

It’s possible that modern AI tools could have a similar impact on evolution, but I’d argue that this is only because computers as a whole will. Large Language Models and their cousins are incremental applications of existing technology rather than something wholly new.

I also think it’s important to not be carried away by the hype driven by AI companies themselves. Bloomberg earlier this year:

Now, a sea of companies are adding “AI” to their taglines and pitch decks, seeking to bask in the reflected glow of the hype cycle. For example, one startup that offers tools to zhuzh up PowerPoints said in a press release that it will incorporate AI so users can skip the writer’s block and build compelling presentations. (It made no mention of AI in a press release describing the product earlier in the year.) Another release touted the value of AI in a campaign to promote shoes.

Perhaps it’s best not to read too much into the marketing. This is a phase change for the tech industry, but I don’t think it’s one for human civilization.

I think it’s important to understand that what’s happening today in social media is not because decentralization’s time has come: it’s because Twitter’s time has gone.

The core need being expressed by millions of users isn’t “get me a decentralized protocol that nobody owns where I can have my choice of algorithms and apps”. It’s “get me a platform that works consistently, with less abuse”. Sometimes it’s also appended with, “where I can build a following for me / my brand / my employer and measure my progress.”

Now, of course, as product people, we can build that with decentralized tech, which will in turn yield benefits later on as an ecosystem grows around it. And third-party app developers probably do want the assurances of an open platform. But most people do not have a nuanced view on how social media is built or how it should be governed. They know what they want for themselves.

“We’ve built the infrastructure for nuanced moderation” is not an adequate answer to people who are suffering, or who are prone to suffer, abuse. “We are making sure you have a safe space to be social” is the only answer for them. If those measures happen to work as part of a nuanced decentralized protocol, great. But either way, it’s got to happen, and it’s got to be at least as good as it would be on any other social network.

Many of us have been wanting decentralized social networking for a long time — I’ve been a part of these conversations for around twenty years. It’s tempting to feel like people finally get it. But that’s a trap and a mistake. As always, quite rightly, most people want something that works for them. If decentralized tech gets them there better than the alternative (and I think it can!) then there’s a wonderful route forward for everyone. But decentralization is not the goal. The goal is always a human experience for people who do not and should not care how the sausage gets made.

I’m soliciting prompts for discussion. This piece is a part of that series.

An anonymous reader asks:

‌Why do emails in jail cost 50 cents each?

Since around 2016, US inmates in 43 state prison systems and the Federal Bureau of Prisons have been able to send electronic messages of a sort. Recipients must be whitelisted and all messages are screened. The upper-bound cost is around 50 cents; the average is now around 30 cents. In Connecticut, they’re free.

To be clear, while these messages are electronic, they’re not email, and are not sent via email protocols. The messages themselves have character limits and don’t support attachments; they also typically don’t support the Unicode character sets that would allow many non-English languages to be used. Inmates are often charged per minute to use the tablets that allow them to send and receive mail to begin with. Data about the messages and their recipients is mined and shared with the prison authority.

Perhaps unsurprisingly, this is an unregulated industry. The proceeds of these fees are shared with the prisons. The tablets themselves are often made freely available to prisons, because each service on them makes their supplier money. Because the revenue is split with the prisons, most prison systems are not going to complain about the unfair pricing policies.

One of the most popular providers, JPay, is largely a prison financial services company that was fined under the Consumer Financial Protection Act for gouging its customers. You may recall a John Oliver piece on prison labor that included a mention of its 45% transfer fees some years ago. Some prisons were contractually obligated to phase out in-person visits, meaning that video calls and these electronic messages were the only way many inmates could contact their families, sending them into debt in the process.

Debt is an important component of this exploitation. The real underlying reason for these high costs is a combination of collusion between prison authorities and the private providers, and a lack of concern for prisoners and their families. Financial debt has a negative impact on re-entry outcomes, as well as family well-being and mental health. Or to put it another way, it keeps people in the prison system, where they can be exploited for labor to the tune of $11bn worth of goods and services a year. With this amount of money on the line, it’s in nobody’s interest to create equitable conditions for prisoners.

Elsewhere in the world, prisons are experimenting with internet access in prisons without this exploitative profit motive. In parts of Australia, Germany, the Philippines, and elsewhere, access with varying degrees of supervision is allowed. This corresponds to a different attitude to incarceration: one that centers around rehabilitation and preventing re-entry rather than imprisonment for profit. Given the wildly disproportionate degree to which people of color are incarcerated in the United States, it’s not too hard to determine the underlying cause of this difference in attitudes.

The problem has more to do with a societal approach; the corruption of individual prison systems and service providers is an outcome of this rather than the cause. In America, rather than people who often need help, the incarcerated are resources to be exploited, at the center of a system that keeps almost two million people behind bars. And that’s why we charge them to message their loved ones.

I’m soliciting prompts for discussion. This piece is a part of that series.

Hunter Walk asks, in two prompts:

What's one business model you wish consumers would get more comfortable with because it would be a net-positive for media/news/society? And similarly, what's one business model you wish media/news organization would spend more time figuring out because it would be net-positive for society?

Should America have more BBC-like funded news media? Is capitalism, democracy and stability in media just fundamentally incompatible for the US?

I’ve conflated these into one piece because I think these are interrelated questions that need to be considered together. How can news organizations find sustainability while keeping the voting public well-informed and retaining their editorial independence?

Let’s begin with the BBC, which is often held up as a model for both newsroom operations and revenue.

The BBC, of course, has traditionally been predominantly broadcast news: a term that will eventually fade from relevance as broadcast television and radio are replaced by the internet over the next decade. It was funded through the UK’s broadcast receiver tax 119 years ago, originally created for radio; as televisions also used broadcast receivers, it was expanded to that equipment when TV came online in 1936. The license fee is essentially a regressive tax which now charges households £159 (roughly $200) a year for the right to use at least one broadcast receiver.

There is no equivalent internet connection tax. Instead, assuming that a household doesn’t own a broadcast receiver, they must continue to pay the license fee if they watch or listen to any kind of live television — for example, on the BBC’s iPlayer service, on YouTube, and so on, regardless of the device they watch it on. This revenue goes to pay for the BBC’s domestic activities in the UK, while its international activities are supported by advertising and traditional licensing through BBC Worldwide, a private corporation set up to drive revenue for the public corporation.

There’s a lesser-known second public broadcaster in the UK: Channel 4 is publicly-owned but makes its revenue through commercial means. It is banned from making its own shows, and must instead re-invest its revenue into distinctive programming as a way to stimulate the creative sector in the UK. Its news program, Channel 4 News, produced by the fully-private news agency ITN, is anecdotally at least as good as the BBC’s: it doggedly broke a number of political scandals in the UK as well as the Cambridge Analytica story. (Overall, the BBC’s coverage is undeniably more expansive, across many broadcast and digital channels.)

Both entities are theoretically independent, but have experienced a fair amount of political interference over the last few years. The BBC has been a political football at the hands of the Conservative Party, which has often felt it has been too critical of its policies. Channel 4 was set to be fully-privatized by that same government, but those plans were abandoned at the beginning of this year.

In the US, the closest analogy is the Corporation for Public Broadasting, a non-profit which receives money from Congress every year and whose board is appointed by the President of the United States. It is legally required to maintain “objectivity and balance”. In turn, the CPB funds stations affiliated with NPR (a national radio non-profit) and PBS (its television equivalent) through both content grants and community service grants. In both cases, the rest of the funding is derived from donations from individuals and private foundations.

The CPB, too, has been a political football. During the most recent war in Iraq, its leadership decided that coverage was too liberal; in response, PBS aired a show headlined by Tucker Carlson for a year. That particular chairman, Kenneth Tomlinson, who had been appointed by George W Bush, was eventually forced to step down after he was found to be directly interfering with programming.

The CPB has limits. There are many non-profit newsrooms in the US that aren’t affiliated with PBS or NPR and don’t broadcast in the traditional sense. These are typically funded through a combination of foundations, gifts from high net worth individuals, and small donations from ordinary people, in reverse order of prominence.

In these cases, one source can’t exist without the others: although foundations or high net worth individuals could hold up a newsroom on their own, that newsroom would then effectively be in their pocket (or at least, they would be perceived to be). By soliciting a range of donations from different sources, non-profits can help ensure their independence both effectively and reputationally. Short of an Act of Congress, the more a newsroom is dependent on a single source of funding, the more it is likely to bend to that source’s will. To future-proof its existence — and secure a reputation for independence — a non-profit newsroom must have diversity in funding.

At the same time, public organizations like the CPB, private foundations, and mandated public-private broadcasters like the BBC are vital to maintaining a thriving news ecosystem. Consider a hypothetical newsroom serving a low-income population in what would otherwise be a news desert. In a community where people are having trouble putting food on the table, there are unlikely to be enough individual donations to a newsroom to keep the lights on. On the other hand, we know that in communities without local newsrooms, voter participation decreases and corruption increases, potentially degrading quality of life in that community even further. Corruption in one area can also lead to corruption in others, creating a larger risk to democracy overall. While this hypothetical community might not present a great profit opportunity and may not be able to financially support a newsroom in its own right, ensuring it has coverage is not unimportant.

Meanwhile, for-profit news is lagging. Buzzfeed just shut down its newsroom; well-funded newcomer The Messenger is shaping up to be the Quibi of news; once-valuable properties like Vice are filing for bankruptcy; cable news networks are beginning to look like they will say or do anything to maintain ratings. One growing trend is for for-profit news sites to either convert into non-profits or adopt non-profit business models like philanthropy. In turn, VC funding for news startups is tanking: the odds of a news startup providing the kinds of financial returns that venture capitalists need to see are vanishingly small. Experiments like the blockchain-based Civil were failures bordering on scams. And the advertising industry, at least in the US, is falling off a cliff.

I don’t think American capitalism itself is incompatible with a functioning news ecosystem. There are thousands of domestic newsrooms that, while not necessarily thriving, are at least sustaining. Revenue and audiences overall are growing. But virtually all of the newsrooms I’m talking about are non-profits. While there are outliers — The New York Times is potentially one — for-profit news has generally fared disastrously. Physical newspapers and magazines continue to die, losing 50% of their revenue over the last two decades. Even where digital outlets are not dependent on ads, paywalls prevent most people from consuming their journalism, eroding their mission to begin with — and for most outlets aren’t even that effective at generating revenue. And micropayments are science fiction at best.

There’s been a boom in non-profit local and niche news startups: around 40% of Institute for Nonprofit News members were founded after 2017, and 60% after 2012. Most of this growth has been in local media, where revenue increased by 25% last year, albeit unevenly (remember my hypothetical news desert above). INN attributes some of this to the American Journalism Project, which seems to have had a positive role in stimulating local newsroom growth since it was founded in 2019. Across the non-profit sector, the median growth in revenue for non-profit news outlets was 25% over 4 years. 40-60% of total revenue was from foundations, 30-40% from individual giving, and the last 10% or so from earned revenue sources like advertising.

This giving-based model isn’t actually limited to non-profits, although many foundations will only award grants to non-profit entities. The Lenfest Institute acts as a non-profit arm for for-profit newspapers in Philadelphia. The Guardian is one example of a for-profit newsroom that has made a success of attracting philanthropy, in part by establishing a companion non-profit to attract tax-advantaged donations (not a million miles away from the model Mozilla uses). Even its for-profit arm is held in a trust whose rules dictate that profits must be re-invested into journalism — and has made a great success out of a subscription patronage model that saw recurring donations increase by 87% in three years.

Given all this, how might we ensure a healthy news ecosystem in America?

I think we need to get comfortable with the idea that non-profits do not exist outside of capitalism. At their heart, they’re another kind of corporate entity, with a different set of rules and restrictions.

We also need to get comfortable with the idea that news will be reported by a patchwork of local and niche newsrooms rather than a single branded entity. I’m sure many people will continue to tune into PBS NewsHour, the NBC Nightly News, and Fox, but these will continue to become the equivalent of the Huffington Post, bringing wider attention to other peoples’ reporting — often with a selective bias that has driven erosion in trust in these national institutions for years. Today, the public trusts individual journalists far more than branded newsrooms — a situation that benefits smaller newsrooms that perhaps don’t have brand recognition yet.

I don’t see a world in the near future where the CPB’s remit is expanded to include independent digital newsrooms across the country, although I do wish that we’d get more comfortable with the idea of federal funding, particularly if we accept the idea that journalism is a requirement for a healthy democracy.

Lacking that, I think the 60%/30%/10% model of foundations, individual giving, and earned revenue has shown itself to be relatively robust. For this to work on a larger scale, consumers will need to get comfortable with the idea of paying for news not because access is scarce but because it’s important. Paying to ensure that reporting happens at all — perhaps not even in the place you live — will need to become more normal.

The journalism industry has experimented with lots of different models. At Matter, an accelerator for media startups where I was west coast Director of Investments, we used to precede every demo day with Clay Shirky’s 2009 quote about the declining newspaper industry that “nothing will work, but everything might”. These days, I would say that some things obviously won’t work (venture capital investment in news, for one), and that we shouldn’t seek one prevailing business model for every kind of newsroom. Industry outlets like The Information can make great money from paywalls, for example; local outlets like Mission Local probably wouldn’t.

Nonetheless, I wish the industry would spend time focusing on making individual patronage donations as easy and seamless as possible. How can that 30% become a diversified 80%? There is no dedicated OpenCollective for news: a place where I can find newsrooms to donate to that will also provide non-profit fiscal sponsorship to newsrooms so that smaller outlets don’t need to spend time and money incorporating.

Patronage has been successful for newsrooms despite the software they use to solicit donations, not because of it. Today, if I want to donate to a newsroom, I need to do it on that newsroom’s site, using whatever tool they’ve decided to adopt. If they want to let me manage my subscription in a meaningful way, they often have to build their own member portal. If they want to transparently allow the public to see exactly who has donated, they have to build that functionality themselves. These are organizations that rarely have full-time developers at all; they shouldn’t have to do any of those things. What is shared infrastructure for making patronage as simple as possible across both non-profit and for-profit newsrooms, both for readers and for the newsrooms themselves?

The news isn’t incompatible with American capitalism. But it may be incompatible with 1980s-era models and outdated delivery mechanisms. Newsrooms will need to continue to collaborate and evolve if they are to survive. And we need them to survive more than ever.

I’m going to try something new for the next few weeks. I usually write about the intersection of media, technology, and democracy, often in reaction to new developments or something that’s been in the news.

But I’d love to know what you’re interested in. So here’s the idea: until further notice, I’m taking prompts. For every prompt or question that’s within the scope of this space, I will commit to writing a post in response to it. Consider me your personal analyst - or perhaps, less charitably, your personal ChatGPT - for free. You can submit them here; if you leave your name, I’ll credit you with a link to your personal website.

Some ideas for prompts include:

• What are the biggest genuine risks posed by AI?
• What could save the non-profit news industry?
• How can open source social networks be welcoming places for vulnerable communities?
• If you were founding a startup in 2023, what would you focus on?
• What will be the year of Linux on the desktop?

But I bet you can do better than that.

Send them to me here. Let’s see how this goes.

Small print: links must be to personal websites only; prompts must be in good faith and not promotions; posts will not be revised; I may stop doing this for any reason, for example if this doesn’t work out or if it becomes popular beyond my wildest dreams.

And that's the thing -- the best thing doesn't win -- the one that is easiest for others to build off of is what wins!

~ Christina Warren on Mastodon, discussing ActivityPub and AT Protocol

This is exactly right. It’s not about ideological purity or architectural perfection.

Can a developer take your protocol, without permission or pushback, and build something meaningful in under a day? That’s how you win.

I’ve loved every moment of working with The 19th. I was a supporter before I joined, and I’ll continue to be one afterwards. As well as well-executed journalism at the intersection of gender, politics, and policy, The 19th is a masterclass in building an equitable remote organizational culture that should serve as a model to other newsrooms and startups. (Hopefully, in part thanks to the documentary about it that will be released at this year’s Tribeca Film Festival, it will be.) The CTO role is now open, and I cannot recommend it highly enough.

So what’s next?

The internet is at a new crossroads. Mainstay services that have been constants for over a decade are ceding space to newcomers that challenge prevailing venture capital business models and offer equitable alternatives. The over-financialization of the web is giving way to co-operatives, collectives, and true decentralization.

This is a movement at the heart of the open web that I care very deeply about. We’re in a very unique moment in time where that movement could succeed, empowering communities everywhere - or it could topple back and be replaced by the same old extractive gatekeepers.

I began my career by building the first website for a local paper in my hometown. I co-founded Elgg, an open source social networking platform that was used by Oxfam to train aid workers as well as universities like Harvard and Stanford. I was a part of the indieweb community, helping people to own their own web presence apart from silos like Facebook - and co-founder of the indieweb platform Known, which powered KQED Teach, a site that won an award from the National Association for Media Literacy. I helped train newsrooms in human-centered design at Matter (where I also invested in mission-driven startups) and taught equitable product design workshops at the Newmark J-School.

I want to use these ideas - human-centered design, open software development, and radical collaboration - in service of the next phase of media and the open web. I want to work with organizations that are similarly motivated, and who have empathetic, inclusive remote work cultures.

What does that look like, exactly? I’m not quite sure yet. It could involve joining an organization or it could involve starting one. It could involve advising many projects or concentrating on one. It will certainly involve lots of experiments.

If this mission resonates with you and you have a Ben-shaped hole in your organization - or if you want to help support these kinds of projects and communities - let’s talk. You can always email me at ben@werd.io.

Thanks for sticking with me. Let’s find out what happens next together.

I’m learning that I cannot write at night. Many writers do their best work once everyone else has gone to bed when the house is quiet; I, on the other hand, am a ragged, sorry mess.

This is a bit of a turnaround for me: I wrote the first version of Elgg in the evenings, usually logging off at a little past 1am. But the rigors of parenting an infant have meant that I’ve become a morning person by force.

So right now I’ve really got two options: wake up really early, and write before everyone else wakes up. (After I’ve made my first cup of coffee, obviously.) Or carve out time and write during baby’s first nap, which is usually somewhere between one to two hours. The latter has been working out pretty well for me lately, but I’ve also been booking calls during that slot.

New rule, then, at least while I’m the primary carer for our son (perhaps it’ll change if we start sending him to daycare or hire a nanny). The morning slot is for writing. The afternoon slots can be used for calls. I need to make that first naptime sacrosanct, otherwise I’m never going to finish this thing.

And I’d like to finish this thing.

A quick reminder: if your favorite service has added generative AI to its core functionality, that means it’s almost certainly also added sending your data to an AI service. Depending on which service that is, that may include sending your data across borders and adding personal information to a training corpus.

It’s worth noting that companies like Google internally ban sending sensitive data to AI services. You should too — particularly if you deal with peoples’ personal information. This functionality can seem magical, but it’s not without cost. As with any technology, it’s important to consider the real implications before making it a part of your business.

My post about AI in the newsroom applies to any small organization. And if you have questions about how you might take advantage of the technology, or what the issues might be, I’m here for you. As always, you can send me an email at ben@werd.io.

Fiction writers are popularly split into two camps: plotters and pantsters. Whereas plotters work closely on a detailed outline before they ever begin a word, iterating on the plot again and again so that it’s tight and hits the right themes, pantsters have a concept in mind, fill their heads with research and ideas, and then just start writing.

I’ve tried really hard to be a plotter, but try as I might, I’ll always be a pantster: in writing, work, and life. In fact, the more I try to plot and create the perfect plan, the less likely I am to actually start writing and see how it feels. The act of creation involves emotion as well as craft; the more I worry about the perfection of my plan, the more I lose creative momentum. The more I iterate, the more the joy seeps through my fingers, until I’m left with a lifeless skeleton that I don’t have the will to carry on with — and I’m still none the wiser about whether my outline would have ever worked.

Some people have the confidence and internal fortitude to build a plan and stick to it; I do not. I self-question like it’s an Olympic sport. In order to overcome this, I need my internal excitement to outweigh my hesitations. Emotional momentum — the kind of excitement that makes you want to dance on your chair because you love the process of what you’re doing so much — is the only way I can get any work out the door.

Doing work imperfectly requires a different kind of confidence. The actor Richard Kind talks about having the confidence to know you’re good at what you do. You can’t just think it speculatively; you’ve got to know it, which means (if you’re anything like me) you’ve got to trick yourself into knowing it.

There are two things I couldn’t have done my first startup without. The first is universal healthcare. (Entrepreneurship is entirely a more brutal proposition without a social safety net.) The second is absolute blind naïvety. If I’d known what I was doing in any way, there’s no way I would have done it. But because I didn’t know enough to ask some of the questions I should have considered, I did it, and it worked. Instead, when a problem arose, we found a way around it, often from first principles.

It’s not that being naïve magically made the problems go away; it’s that we had emotional and intellectual momentum, and we had the confidence to know that we could overcome problems that arose. We weren’t blind: we had a North Star, knew broadly what we were trying to achieve, and had a good understanding of the people we were building for. But we weren’t dead set on doing it a particular way. We kept an open mind. And that’s how we ended up building software that was originally built for higher education but found use at organizations like Oxfam, in social movements like the Spanish 15-M anti-austerity movement, and at Fortune 500 companies. We didn’t know any of that was going to happen ahead of time, but we scrappily adapted and grew. We were pantsters.

I’m trying hard to finish a novel, and do it seriously. It’s hard work, and although there are some similarities to finishing any large creative project, the craft involved is very different to building software. I’m also a very different person to the naïve kid who built a social network twenty years ago. For one thing, I don’t have anywhere near as much free time. For another, my self-doubt is so much better informed.

It’s taken me too long to realize that I have to work on is that emotional momentum. At least for the first draft. It’s not the only thing, and I’m prepared to work hard chiseling whatever comes out into something palatable. But first, the excitement, the creative flow, the momentum, the force.

And when I build that next big software project from scratch, I’ll have to re-learn it then, too.

I’m disproportionately frustrated by news websites that don’t provide an RSS feed. Sure, most provide an email newsletter these days, and that will suit many users. (It also suits the publisher just fine, because now they know exactly who is subscribing.) But while it’s been around for a long time, RSS isn’t the niche technology many people seem to think it is.

I start every day by reading my feeds in Reeder: a popular way for Apple users to keep on top of new content from their favorite sites. There are plenty of alternatives for every platform you can think of. On top of all the easy-to-use open news readers that are available, apps like Apple News also use a dialect of RSS behind the scenes. It is the standard way for websites to let people read updates.

It’s also a way for publishers to free themselves just a little bit more from the proprietary social media ecosystem. If most users learn about content they’re interested in from Facebook, publishers are beholden to Facebook. If most users learn about new stories from open web standards like RSS, publishers aren’t beholden to anybody. They have full control — no engagement from the partnerships team required.

It’s very cheap to support. If you’re using a CMS like WordPress, it comes free out of the box; there’s no email inbox to clog up; and not allowing people to subscribe directly is hostile to both the user and the publisher. Hell, if you really want to, you can even run ads in the feed.

So, please: I want to read your articles. Spend half a day of developer time and set up a feed for every site you run.

Thank you.

Ryan discusses the differences between the fediverse and the AT Protocol:

One core difference between the fediverse and the AT Protocol seems to be that AT decouples many key building blocks – identity, moderation, ranking algorithms, even your own data to some degree – from your server. The fediverse, on the other hand, ties them all to your server and sees that as a desirable feature.

I’m probably being a bit presumptuous, but I think there’s actually a difference between a European and American mindset here. (Mastodon is headquartered in Germany while Bluesky is rooted in San Francisco and Austin.)

The fediverse prioritizes communities: each community instance has its own rules, culture, and potentially user interface. You find a community that you’re aligned with first and foremost, and your activity is dictated by that.

The AT Protocol is much more individualistic. You bring your own identity support, moderation, ranking algorithms, interface, etc. You’re using someone’s space to be able to access the network, but ultimately your choices are yours rather than an outcome of which collaborative community you’ve opted to join.

I think both models are good. I like the fediverse’s emphasis on community. I also think by not emphasizing granular community rules early on, Bluesky has the luxury of being able to build community across the whole network more cohesively. I’m glad both exist.

Google is adding to Brand Indicators for Message Identification:

Building upon that feature, users will now see a checkmark icon for senders that have adopted BIMI. This will help users identify messages from legitimate senders versus impersonators.

So in other words, Gmail will show a blue checkmark for email domains that have logged a registered trademark, bought a Verified Mark Certificate, and set up DMARC.

I hate this!

Although this method avoids Google itself from being a central authority, it demands that senders (1) have a verifiable registered trademark, (2) pay well over a thousand dollars for a Verified Mark Certificate.

This heavily disadvantages small vendors, sole operators, and anyone who can’t afford to drop a couple of thousand dollars on their email domain. The effect is to create an aura of legitimacy for larger organizations at the expense of individuals and smaller shops. It also heavily advantages certificate vendors, who are already running what amounts to be shakedown scam across the whole internet.

It’s an unequal, annoying policy, made worse by the realization that Gmail is likely to add routing rules that advantage BIMI-enabled messages in the future. Bah, humbug.

I’m focusing on the intersection of technology, media, and democracy. Subscribe by email to get every update.

It’s genuinely refreshing to see how non-profit newsrooms have been embracing the open web and the spirit of collaboration over competition. These are often resource-strapped organizations shedding light on underreported stories, many of which are local or apply to vulnerable communities. They’re usually donation-supported rather than paywalled, and the primary goal is to get the journalism out and serve the public. They’re public service organizations first and foremost.

You’ve probably seen newsrooms like The 19th, ProPublica, Grist, and The Texas Tribune. What you might not have noticed is that each of them makes their articles available under an Attribution NonCommercial NoDerivatives Creative Commons license, such that anyone can republish them on their own sites. Publisher by publisher, a nascent ecosystem for open news content is being built.

There are a few carve-outs: often photos are not re-licensable, so republishable content usually comes without illustrations. There’s also often an analytics pixel included in the content so that newsrooms can measure their reach and report back to their funders.

And the reach can be significant. By making their content available under an open license, these newsrooms can find audiences far beyond their websites: major outlets like PBS, USA Today, the Washington Post, and more are all actively republishing stories.

Because of the turnaround time involved in one outlet reporting on and publishing a story to their site, and another discovering it, re-illustrating it, and publishing it on their own site, this mechanism hasn’t been particularly applicable to breaking news. But there’s a lot of potential in gathering feeds from open publishers and creating a kind of republishing newswire, which could speed up this process and streamline the ability for these newsrooms to reach other publishers and audiences.

Grist just announced Rural Newswire, which is exactly that: a collection of publishers reporting on rural America that make their content available under a Creative Commons license. The site contains a filterable, RSS-powered feed with “republish” buttons next to each story. It’s the first site like this I think I’ve seen, but I know more are coming — and, of course, there’s nothing stopping third parties from creating their own. Each RSS feed is publicly available and instructions for republishing are provided by each site.

The result is a de facto co-operative of non-profit news organizations, working together to build a commons that makes the country more informed. It’s a way that open content licensing and open source ideas are really working to strengthen democracy. It’s the kind of thing that gives me hope for the future.

I’m focusing on the intersection of technology, media, and democracy. Subscribe by email to get every update.

By now, you’ve been exposed to Generative AI and Large Language Models (LLMs) like OpenAI’s ChatGPT, DALL-E 2, and GPT-4. It seems a lot like magic: a bot that seems to speak like a human being, provides confident-sounding answers, and can even write poetry if you ask it to. As an advance, it’s been compared in significance to the advent of the web: a complete paradigm shift of the kind that comes along very rarely.

I want to examine their development through the lens of a non-profit newsroom: specifically, I’d like to consider how newsrooms might think about LLMs like ChatGPT, both as a topic at the center of reporting, as well as a technology that presents dangers and opportunities internally.

I’ve picked newsrooms because that’s the area I’m particularly interested in, but also because they’re a useful analogue: technology-dependent organizations that need to move quickly but haven’t always turned technology into a first-class competence. In other words, if you’re not a member of a non-profit newsroom, you might still find this discussion useful.

What are generative AI and Large Language Models?

Generative AI is just an umbrella term for algorithms that have the ability to create new content. The ones receiving attention right now are mostly Large Language Models: probability engines that are trained to predict the next word in a sentence based on a very large corpus of written information that has often been scraped from the web.

That’s important to understand because when we think of artificial intelligence, we often think of android characters from science fiction movies: HAL 9000, perhaps, or the Terminator. Those stories have trained us to believe that artificial intelligence can reason like a human. But LLMs are much more like someone put the autocomplete function on your phone on steroids. Although their probabilistic models generate plausible answers that often look like real intelligence, the algorithms have no understanding of what they’re saying and are incapable of reasoning. Just as autocomplete on your phone sometimes gets it amazingly wrong, LLM agents will sometimes reply with information that sounds right but is entirely fictional. For example, the Guardian recently discovered that ChatGPT makes up entire news articles.

It’s also worth understanding because of the provenance of the datasets behind those models. My website — which at the time of writing does not license its content to be re-used — is among the sites scraped to join the corpus; if you have a site, it may well be too. There’s some informed conjecture that these scraped sites are joined by pirated books and more. Because LLMs make probabilistic decisions based on these corpuses, in many ways their apparent intelligence could be said to be derived from this unlicensed material. There’s no guarantee that an LLM’s outputs won’t contain sections that are directly identifiable as copyrighted material.

This data has often been labeled and processed by low-paid workers in emerging nations. For example, African content moderators just voted to unionize in Nairobi.

Finally, existing biases that are prevalent in the corpus will be reiterated by the agent. In a world where people of color are disproportionately targeted by police, it’s dangerous to use an advanced form of autocomplete to determine who might be guilty of a crime — particularly as a software agent might be more likely to be incorrectly assumed to be impartial. As any science fiction fan will tell you, robots are supposed to be logical entities who are free from bias; in reality they’re only as unbiased as their underlying data and algorithms.

In other words, content produced by generative AI may look great but is likely to be deeply, sometimes dangerously flawed.

Practically, the way one interacts with them is different to most software systems: whereas a standard system might have a user interface with defined controls, a command line argument structure, or an API, you interact with an LLM agent through a natural language prompt. Prompt engineering is an emergent field.

Should I use LLMs to generate content?

At the beginning of this year, it emerged that CNET had been using generative AI to write whole articles. It was a disaster: riddled with factual errors and plodding, mediocre writing.

WIRED has published a transparent primer on how it will be using the technology.

From the text:

The current AI tools are prone to both errors and bias, and often produce dull, unoriginal writing. In addition, we think someone who writes for a living needs to constantly be thinking about the best way to express complex ideas in their own words. Finally, an AI tool may inadvertently plagiarize someone else’s words. If a writer uses it to create text for publication without a disclosure, we’ll treat that as tantamount to plagiarism.

For all the reasons stated above, using AI to generate articles from scratch, or to write passages inside a published article otherwise written by a human, is not likely to be a good idea.

The people who will use AI to generate articles won’t surprise you: spammers will be all over it as a way to cheaply generate clickbait content without having to hire writers. The web will be saturated with this kind of low-quality, machine-written content — which means that it will be incumbent on search engines like Google to filter it out. Well-written, informative, high-quality writing will rise to the top.

There’s another danger, too, for people who are tempted to use LLMs to power chat-based experiences, or to use them to process user-generated content. Because LLM agents use natural language prompts with little distinction between the prompt and the data the LLM is acting on, prompt injection attacks are becoming a serious risk.

And they’re hard to mitigate. As Simon Willison points out in the above link:

To date, I have not yet seen a robust defense against this vulnerability which is guaranteed to work 100% of the time. If you’ve found one, congratulations: you’ve made an impressive breakthrough in the field of LLM research and you will be widely celebrated for it when you share it with the world!

Finally, let’s not forget that unless you’re running an LLM on your own infrastructure, all your prompts and outputs are being saved on a centralized service where your data almost certainly will be used for further training the model. There is little to no expectation of privacy here (although some models are beginning to offer enterprise subscriptions that promise but don’t demonstrate data privacy).

Then what can I use LLMs for?

Just as autocomplete can be really useful even if you’d never use it to write a whole essay that you’d show to anyone else, LLMs have lots of internal uses. You can think of them as software helpers that add to your process and potentially speed you up, rather than a robot that will take your job tomorrow. Because they’re helping you build human-written content rather than you publishing their machine-written output, you’re not at risk of violating someone’s copyright or putting a falsehood out into the world unchecked. Prompt injection attacks are less hazardous, assuming you trust your team and don’t expose agents to unchecked user-generated content.

Some suggestions for how LLMs can be used in journalism include:

• Suggesting headlines
• Speeding up transformations between media (for example, articles to short explainers, or to scripts for a video)
• Automatic transcription from audio or video into readable notes (arguably the most prevalent existing use of AI in newsrooms)
• Extracting topics (that can then be linked to topic archive pages)
• Discovering references to funders that must be declared
• Suggesting ideas for further reporting
• Uncovering patterns in data provided by a source
• Community sentiment analysis
• Summarizing large documents

All of these processes can sit within a content management system or toolset as just another editing tool. They don’t do away with the journalist or editor: they simply provide another tool to help them to do their work. In many cases they can be built as CMS add-ons like WordPress plugins.

Hosting is another matter. When newsrooms receive sensitive leaks or information from sources, interrogating that data with a commercial, centrally-hosted LLM may not be advisable: doing so would reveal that sensitive data to the service provider. Instead, newsrooms likely to receive this kind of information would be better placed to run their own internal service on their own infrastructure. This is potentially expensive, but it also carries another advantage: advanced newsrooms may also be able to build and train their own corpus of training data rather than using more generic models.

Will LLMs be a part of the newsroom?

Of course — but beware of the hype machine. This kind of AI is a step forward in computing, but it is not a replacement for what we already use. Nor is it going to be the job-destroyer or civilization-changer some have predicted it to be (including VCs, who currently have a lot to lose if AI doesn’t live up to its frothily declared potential).

It’s another creative ingredient. A building block; an accelerator. It’s just as if — imagine that — autocomplete was put on steroids. That’s not nothing, but it’s not everything, either. There will be plenty of really interesting tools designed to help newsrooms do more with scant resources, but I confidently predict that human journalists and editors will still be at the center of it all, doing what they do best. They’ll be reporting, with a human eye — only faster.