2026-02-04 08:00:00
This past week, over a series of events, the New York tech community celebrated the 30th anniversary of a nebulous idea described as “Silicon Alley”, the catch-all term for our greater collective of creators and collaborators, founders and funders, inventors and investors, educators and entrepreneurs and electeds, activists and architects and artists. Some of the parties or mixers have been typical industry affairs, the usual glad-handing about deal-making and pleasantries. But a lot have been deeper, reflecting on what’s special and meaningful about the community we’ve built in New York. Steven Rosenbaum’s reflection on the anniversary captures this well from someone who’s been there, and Leo Schwartz’s piece for Fortune covers the more conventional business angle.
Beyond the celebrations, though, I wanted to reflect on a number of the deeper conversations I’ve had over these last few days. These are conversations grounded in the reality of where our country and city are today, far beyond spaces where wealthy techies are going to parties and celebrating each other. The hard questions raised in these conversations are the ones that determine where this community goes in the future, and they’re the ones that every tech community is going to face in the current moment.
I know what the New York City tech community has been; there was a time when I was one of its most prominent voices. The question now is what it will be in the future. Because we are at a profound crossroads.
Nobody better exemplifies the best of what New York tech has been than Aaron Swartz. As I’d written about recently, he was brilliant and delightfully impossible. At an incredibly young age, he led our community in the battle to push back against a pair of ill-considered bills that threatened free expression on the Internet. (These bills would have done to the web what the current administration has done to broadcast television, having a chilling effect on free speech and putting large swaths of content under government control.) As we stood outside Chuck Schumer’s office and demanded that big business take their hands off our internet, we got our first glimpse of the immense power that our community could wield. And we won, at least for a while.
My own path within the New York tech community was nowhere near as dramatic, but I was just as motivated in wanting to serve the community. When I became the first person elected to the board of the New York Tech Meetup (later the New York Tech Alliance), it was the largest member-led organization of tech industry workers in the country. By the time it reached its peak, we were over 100,000 members strong, and could sell out one of our monthly events (at a venue of over 1000 attendees) in minutes. The collective power and impact of that cohort was immense. So, when I say “community”, I mean community. I’m not talking about the contemporary usage of the word, when people call their TikTok followers a “community”. I mean people who care about each other and show up for each other so that they can achieve meaningful things.
New York tech demonstrated its values time and again, and not just in organizing around policy that served its self-interest. When the city was still reeling from 9/11, these were people who not only chose to stay in the city, or who simply talked about how New York ought to rebuild, but actually took the risk and rebuilt the economy of the city — the majority of the economic regrowth and new jobs in New York City in the quarter-century since the attacks of 9/11 have happened thanks to the technology sector.
When Hurricane Sandy hit, these were people who were amongst the first to step up to help their neighbors dig out. When our city began to open up its data, the community responded in kind by building an entire ecosystem of new tools that laid the groundwork for the tech we now take for granted when navigating around our neighborhoods. There was no reluctance to talk about the importance of diversity and inclusion, and no apology in saying that tech was failing to do its job in hiring and promoting equitably, because we know how much talent is available in our city. Hackers would come to meetups to show off their startups, sure, but just as often to show off how they’d built cool new technology to help make sure our neighbors in public housing had heat in the winter. This was New York-style tech.
What’s more, the work of this community happened with remarkable solidarity; the SOPA/PIPA protests that Aaron Swartz spoke at had him standing next to some of the most powerful venture capitalists in the city. When it was time to take action, a number of the most influential tech CEOs in New York took Amtrak down to Washington, D.C. to talk to elected officials and their staffers about the importance of defending free expression online, advocating for the same issue that had been so important to the broke college kids who’d been at the rally just a few days earlier. People had actually gathered around principles. I don’t say this as a Pollyanna who thinks everything was perfect, or that things would have always stayed so idealistically aligned, but simply to point out that this did happen. I don’t have to assert that it is theoretically possible, because I have already seen a community which functions in this way.
But things have changed in recent years for New York’s tech community. What used to often be about extending a hand to neighbors has, much of the time, become about simply focusing on who’s getting funded to chase the trends defined by Silicon Valley. The vibrancy of the New York Tech Meetup took a huge hit from covid, preventing the ability for the community to gather in person, and the organization’s evolution from a Meetup to an Alliance to being part of Civic Hall shifted its focus in recent years, though there has been a recent push to revitalize its signature events. In its place, much of the public narrative for the community is led by Tech:NYC, which has active and able leadership, but is a far more conventional trade group. There's a focus on pragmatic tools like job listings (their email newsletter is excellent), but they're unlikely to lead a rally in front of a Senator's office. An organization whose founding members include Google and Meta is necessarily going to be different than one with 100,000 individual members.
When I spoke to the Wall Street Journal back in 2013 about the political and social power of our community, at a far different time, I called out the breadth of who our community includes:
The tech constituency encompasses a range of potential voters who remain unlikely to behave as a traditional bloc. "It's venture capitalists and 23-year-old graphic designers in Bushwick," Mr. Dash said. "It's labor and management. It's not traditional allies."
I wanted to make sure people understood that tech in New York is much broader than just, well, what the bosses and the big companies want. It is important to understand that New York is about founders, not just funders.
The distinction between these groups and their goals was never clearer to me than in the 2017 battle around Amazon’s proposed HQ2 headquarters. The public narrative was that Amazon was trying to make a few cities jump through hoops to make the best possible set of bribes to the company so that they would build a new headquarters complex in the host city. The reality was, New York City offered $1.5 billion dollars to the richest man in the world in order to open up an office in a city where the company was inevitably going to do business regardless, and the contract that Amazon would have to sign in exchange only obligated them to hire 500 new workers in the city — fewer people than their typical hiring plan would expect in that timeframe. In addition, the proposed plan would have taken over land intended for 6,000 homes, including 1500 affordable units, would have defunded the mass transit system through years of tax breaks for the company while putting massive additional burden on the transit system, and raised housing prices. (Amazon has since signed a lease for 335,000 square feet and hired over 1000 employees, without any subsidies.)
At the time, I was CEO of a company that two entrepreneurs had founded in 2000 and bootstrapped to success, leading to them spinning out multiple companies which would go on to exit for over $2.2 billion, providing over 500 jobs and creating dozens of millionaires out of the workers who joined the companies over the years. Several of the people who had worked at those companies went on to form their own companies, and those companies are now collectively worth over $5 billion. All of these companies, combined, have gotten a total of zero billion dollars from the state and city of New York. In addition, none of those companies have ever had working conditions anywhere close to those Amazon has been criticized for.
But the story of the time was that “New York tech wants HQ2!” Media like newspapers and TV were firmly convinced that techies were in support of Amazon getting a massive unnecessary handout, and I had genuinely struggled to figure out why for a long time. After a while, it became obvious. Everyone that they had spoken to, and all the voices that were considered canonical and credible when talking about “New York tech”, were investors or giant publicly-traded companies.
People who actually built things were no longer the voice of the community. Those who showed up when the power was out, or when the community was hurting, or when there was an issue that called for someone to bravely stand up and lead the crowd even if there was some social or political risk — they were not considered valid. People liked the myth of Aaron Swartz by then, but they would have ignored the fact that he almost certainly would have objected to corporate subsidy for the company.
I am still proud of the New York tech community. But that’s because I get to see what happens in person. Last week, I was reminded at every one of the in-person commemorations of the community that there are so many generous, kind-hearted, thoughtful people who will fight to do the right thing. The challenge today, though, is that those are no longer the people who define the story of the community. That’s not who a new person thinks of when they’re introduced to our community.
When I talk to young people who are new to the industry, or people who are changing careers who are curious about tech, they have heard of things like Tech Week, or they read trade press. In those venues, a big name is generally not our home-grown founders, or even the “big” success stories of New York tech. That’s especially true as once high-flying New York tech companies like Tumblr and Foursquare and Kickstarter and Etsy and Buzzfeed either faded or got acquired, and newer successful startups are more prosaic and less attention-grabbing. Who’s left to tell them a story of what “tech” means in New York? Where will they find community?
One possible future is that they try to build a startup, doing everything you’re “supposed” to do. They pitch the VC firms in town, and the big name firms that they’ve heard of. If they’re looking for community, they go to the events that get the most promotion, which might be Tech Week events. And all of these paths lead the same way — the most prominent VC firm is Andreessen Horowitz, and they run Tech Week too, even though they’re not from NYC.
On that path, New York tech puts you across the table from the man who strangled my neighbor to death.
Another possible future is that we rebuild the kind of community that we used to have. We start to get together the people who actually make things, and show off what we’ve built for one another. It’s going to require re-centering the hundreds of thousands of people who create and invent, rather than the dozens of people who write checks. It’s going to mean that the stories start with New York City (and maybe even… in the outer boroughs!), rather than taking dictation from those in Silicon Valley who hate our city. And it’s going to require understanding that technology is a set of tools and tactics we can use in service of goals — ideally positive social goals — and not just an economic opportunity to be extracted from.
We would never talk about education by only talking to those who invest in making pencils. We’d never consider a story about a new movie to be complete if we only talked to those who funded the film. And certainly our policymakers would balk if we skipped speaking with them and instead aimed our policy questions directly at their financial backers, though that might result in more accurate responses. Yet somehow, with technology, we’ve given over the narrative entirely to the money men.
In New York, we’ve borne the brunt of that error. A tech community with heart and soul is in danger of being snuffed out by those who will only let its most base instincts survive. Even our investors here are more thoughtful than these stories would make it seem! But we can change it, and maybe even change the larger tech story, if we’re diligent in never letting the bad actors control the narrative of what tech is in the world.
Like so many good things, it can all start with New York City.
2026-01-27 08:00:00
There's been a remarkable leap forward in the ability to orchestrate coding bots, making it possible for ordinary creators to command dozens of AI bots to build software without ever having to directly touch code. The implications of this kind of evolution are potentially extraordinary, as outlined in that first set of notes about what we could call "codeless" software. But now it's worth looking at the larger ecosystem to understand where all of this might be headed.
One idea that's come up in a host of different conversations around codeless software, both from supporters and skeptics, is how these new orchestration tools can enable coders to control coding bots that aren't from the Big AI companies. Skeptics say, "won't everyone just use Claude Code, since that's the best coding bot?"
The response that comes up is one that I keep articulating as "frontier minus six", meaning the idea that many of the open source or open-weight AI models are often delivering results at a level equivalent to where frontier AI models were six months ago. Or, sometimes, where they were 9 months or a year ago. In any of these cases, these are still damn good results! These levels of performance are not merely acceptable, they are results that we were amazed by just months ago, and are more than serviceable for a large number of use cases — especially if those use cases can be run locally, at low cost, with lower power usage, without having to pay any vendor, and in environments where one can inspect what's happening with security and privacy.
When we consider that a frontier-minus-six fleet of bots can often run on cheap commodity hardware (instead of the latest, most costly, hard-to-get Nvidia GPUs) and we still have the backup option of escalating workloads to the paid services if and when a task is too challenging for them to complete, it seems inevitable that this will be part of the mix in future codeless implementations.
The most thoughtful and fluent analysis of the new codeless approach has been this wonderful essay by Maggie Appleton, whose writing is always incisive and insightful. This one's a must-read! Speaking of Gas Town (Steve Yegge's signature orchestration tool, which has catalyzed much of the codeless revolution), Maggie captures the ethos of the entire space:
We should take Yegge’s creation seriously not because it’s a serious, working tool for today’s developers (it isn’t). But because it’s a good piece of speculative design fiction that asks provocative questions and reveals the shape of constraints we’ll face as agentic coding systems mature and grow.
Once you've considered Maggie's piece, it's worth reading over Steve Krouse's essay, "Vibe code is legacy code". Steve and his team build the delightful val town, an incredibly accessible coding community that strikes a very careful balance between enabling coding and enabling AI assistance without overwriting the human, creative aspects of building with code. In many ways (including its aesthetic), it is the closest thing I've seen to a spiritual successor to the work we'd done for many years with Glitch, so it's no surprise that Steve would have a good intuition about the human relationship to creating with code.
There's an interesting point, however to the core point Steve makes about the disposability of vibe-coded (or AI-generated) code: all code is disposable. Every single line of code I wrote during the many years I was a professional developer has since been discarded. And it's not just because I was a singularly terrible coder; this is often the normal thing that happens with code bases after just a short period of time. As much as we lament the longevity of legacy code bases, or the impossibility of fixing some stubborn old systems based on dusty old languages, it's also very frequently the case that people happily rip out massive chunks of code that people toiled over for months or years and then discard it all without any sentimentality whatsoever.
Codeless tooling just happens to embrace this ephemerality and treat it as a feature instead of a bug. That kind of inversion of assumptions often leads to interesting innovations.
As I noted in my original piece on codeless software, we can expect any successful way of building software to be appropriated by companies that want to profiteer off of the technology, especially enterprise companies. This new realm is no different. Because these codeless orchestration systems have been percolating for some time, we've seen some of these efforts pop up already.
For example, the team at Every, which consults and builds tools around AI for businesses, calls a lot of these approaches compound engineering when their team uses them to create software. This name seems fine, and it's good to see that they maintain the ability to switch between models easily, even if they currently prefer Claude's Opus 4.5 for most of their work. The focus on planning and thinking through the end product holistically is a particularly important point to emphasize, and will be key to this approach succeeding as new organizations adopt it.
But where I'd quibble with some of what they've explained is the focus on tying the work to individual vendors. Those concerns should be abstracted away by those who are implementing the infrastructure, as much as possible. It's a bit like ensuring that most individual coders don't have to know exactly which optimizations a compiler is making when it targets a particular CPU architecture. Building that muscle where the specifics of different AI vendors become less important will help move the industry forward towards reducing platforms costs — and more importantly, empowering coders to make choices based on their priorities, not those of the AI platforms or their bosses.
A good example of the "normal" developer ecosystem recognizing the groundswell around codeless workflows and moving quickly to integrate with them is the Tailscale team already shipping Aperture. While this initial release is focused on routine tasks like managing API keys, it's really easy to see how the ability to manage gateways and usage into a heterogeneous mix of coding agents will start to enable, and encourage, adoption of new coding agents. (Especially if those "frontier-minus-six" scenarios start to take off.)
I've been on the record for years about being bullish on Tailscale, and nimbleness like this is a big reason why. That example of seeing where developers are going, and then building tooling to serve them, is always a sign that something is bubbling up that could actually become signficant.
It's still early, but these are the first few signs of a nascent ecosystem that give me more conviction that this whole thing might become real.
2026-01-26 08:00:00
I've been working in and around the technology industry for a long time. Depending on how you count, it's 20 or 30 years. (I first started getting paid to put together PCs with a screwdriver when I was a teenager, but there isn't a good way to list that on LinkedIn.) And as soon as I felt like I was pretty sure that I was going to be able to pay the next month's rent without having to eat ramen noodles for two weeks before it was due, I felt like I'd really made it.
And as soon as you've made it, you owe it to everybody else to help out as much as you can. I don't know how to put it more simply than that. But for maybe the first decade of being in the "startup" world, where everybody was worried about appealing to venture capital investors, or concerned about getting jobs with the big tech companies, I was pretty convinced that one of the things that you couldn't do to help people was to talk about some of the things that were wrong. Especially if the things that were wrong were problems that, when described, might piss off the guys who were in charge of the industry.
But eventually, I got a little bit of power, mostly due to becoming a little bit visible in the industry, and I started to get more comfortable speaking my mind. Then, surprisingly, it turned out that... nothing happened. The sky didn't fall. I didn't get fired from my jobs. I certainly got targeted for harassment by bad actors, but that was largely due to my presence on social media, not simply because of my views. (And also because I tend to take a pretty provocative or antagonistic tone on social media when trying to frame an argument.) It probably helped that, in the workplace, I both tend to act like a normal person and am also generally good at my job.
I point all of this out not to pat myself on the back, or as if any of this is remarkable — it's certainly not — but because it's useful context for the current moment.
I have been around the technology industry, and the larger business world, long enough to have watched the practice of speaking up about moral issues go from completely unthinkable to briefly being given lip service to actively being persecuted both professionally and politically. The campaigns to stamp out issues of conscience amongst working people have vilified caring for others with names ranging from "political correctness" to "radicalism" to "virtue signaling" to "woke" and I'm sure I'm missing many more. This, despite the fact that there have always been thoughtful people in every organization who try to do the right thing; it's impossible to have a group of people of any significant size and not have some who have a shred of decency and humanity within them.
But the technology industry has an incredibly short memory, by design. We're always at the beginning of history, and so many people working in it have never encountered a time before this moment when there's been this kind of brutal backlash from their leaders against common decency. Many have never felt such pressure to tamp down their own impulses to be good to their colleagues, coworkers, collaborators and customers.
I want to encourage everyone who is afraid in this moment to find some comfort and some solace in the fact that we have been here before. Not in exactly this place, but in analogous ones. And also to know that there are many people who are also feeling the same combination of fear or trepidation about speaking up, but a compelling and irrepressible desire to do so. We've shifted the Overton window on what's acceptable multiple times before.
I am, plainly, exhorting you do to speak up about the current political moment and to call for action. There is some risk to this. There is less risk for everyone when more of us speak up.
In the United States, our government is lying to us about an illegal occupation of a major city, which has so far led to multiple deaths of innocents who were murdered by agents of the state. We have video evidence of what happened, and the most senior officials in our country have deliberately, blatantly and unrepentantly lied about what the videos show, while besmirching the good names of the people who were murdered. Just as the administration's most senior officials spread these lies, several of the most powerful and influential executives in the tech industry voluntarily met with the President, screened a propaganda film made expressly as a bribe for him, and have said nothing about either the murders or the lies about the murders.
These are certainly not the first wrongs by our government. These are not even the first such killings in Minnesota in recent years. But they are a new phase, and this occupation is a new escalation. This degree of lawless authoritarianism is new — tech leaders were not crafting golden ingots to bribe sitting leaders of the United States in the past. Military parades featuring banners bearing the face of Dear Leader, followed by ritual gift-giving in the throne room of the golden palace with the do-nothing failsons and conniving hangers-on of the aging strongman used to be the sort of thing we mocked about failing states, not things we emulated about them.
So, when our "leaders" have failed, and they have, we must become a leaderful community. This, I have a very positive feeling about. I've seen so many people who are willing to step up, to give of themselves, to use their voices. And I have all the patience in the world for those who may not be used to doing those things, because it can be hard to step into those shoes for the first time. If you're unfamiliar or uncomfortable with this work, or if the risk feels a little more scary because you carry the responsibility of caring for those around you, that's okay.
But I've been really heartened to see how many people have responded when I started talking about these ideas on LinkedIn — not usually the bastion of "political" speech. I don't write the usual hustle-bro career advice platitudes there, and instead laid out the argument for why people will need to choose a side, and should choose the side that their heart already knows that they're on. To my surprise, there's been near-universal agreement, even amongst many who don't agree with many of my other views.
It is already clear that business leaders are going to be compelled to speak up. It would be ideal if it is their own workers who lead them towards the words (and actions) that they put out into the world.
Those of us in the technology realm bear a unique responsibility here. It is the tools that we create which enable the surveillance and monitoring that agencies like ICE use to track down and threaten both their targets and those they attempt to intimidate away from holding them accountable. It is the wealth of our industry which isolates the tycoons who run our companies when they make irrational decisions like creating vanity films about the strongman's consort rather than pushing for the massive increase in ICE spending to instead go towards funding all of Section 8 housing, all of CHIP insurance, all school lunches, and 1/3 of all federal spending on K-12 education.
It takes practice to get comfortable using our voices. It takes repetition until leaders know we're not backing down. It takes perseverance until people in power understand they're going to have to act in response to the voices of their workers. But everyone has a voice. Now is your turn to use it.
When we speak, we make it easier for others to do so. When we all speak, we make change inevitable.
2026-01-22 08:00:00
There’s finally been a big leap forward in coding tech unlocked by AI — not just “it’s doing some work for me”, but “we couldn’t do this before”. What’s new are a few smart systems that let coders control fleets of dozens of coding bots, all working in tandem, to swarm over a list of tasks and to deliver entire features, or even entire sets of features, just from a plain-English description of the strategic goal to be accomplished.
This isn’t a tutorial, this is just trying to understand that something cool is happening, and maybe we can figure out what it means, and where it’s going. Lots of new technologies and buzzwords with wacky names like Gas Town and Ralph Wiggum and loops and polecats are getting as much attention as, well, anything since vibe coding. So what’s really going on?
The breakthrough here came from using two familiar ideas in interesting new ways. The first idea is orchestration. Just like cloud computing got massively more powerful when it became routine for coders to be able to control entire fleets of servers, the ability to reliably configure and control entire fleets of coding bots unlocks a much higher scale of capability than any one person could have by chatting with a bot on their own.
The second big idea is resilience. Just like systems got more capable when designers started to assume that components like hard drives would fail, or that networks would lose connection, today’s coders are aware of the worst shortcoming of using LLMs: sometimes they create garbage code. This tendency used to be the biggest shortcoming about using LLMs to create code, but by designing for failure, testing outputs, and iterating rapidly, codeless systems enable a huge advancement in the ultimate reliability of the output code.
The codeless approach also addresses the other huge objection that many coders have to using LLMs for coding. The most common direct objection to using AI tools to assist in coding hasn’t just been the broken code — it’s been the many valid social and ethical concerns around the vendors who build the platforms. But codeless systems are open source, non-commercial, and free to deploy, while making it trivial to swap in alternatives for every part of the stack, including using open source or local options for all or part of the LLM workload. This isn’t software being sold by a Big AI vendor; these are tools being created by independent hackers in the community.
The ultimate result is the ability to create software at scale without directly writing any code, simply by providing strategic direction to a fleet of coding bots. Call it “codeless” software.
If you’re looking for a quick bullet-point summary, here’s something skimmable:
As we’ve been saying for some time, for people who actually make and understand technology, the majority AI view is that LLMs are just useful technologies that have their purposes, but we shouldn’t go overboard with all of the absurd hype. We’re seeing new examples of the deep moral failings and social harms of the Big AI companies every day.
Despite this, coders still haven’t completely written off the potential of LLMs. A big reason why coders are generally more optimistic about AI than writers or photographers is because, in creative spaces, AI smothers the human part of the process. But in coding, AI takes over the drudgery, and lets coders focus on the most human and expressive parts.
The shame, then, is that much of the adoption of AI for coding has been in top-down mandates at companies. Rather than enabling innovation, it’s been in deployments designed to undermine their workers’ job security. And, as we’ve seen, this has worked. It’s no wonder that a lot of the research on enterprise use of AI for coding has shown little to no increase in productivity; obviously productivity improvements have not been the goal, much of the time.
Codeless tech has the potential to change that. Putting the power of orchestrating a fleet of coding bots in the hands of a smart and talented coder (or designer! or product manager! or writer! or…) upends a lot of the hierarchy about who’s able to call the shots on what gets created. The size of your nights-and-weekends project might be a lot bigger, the ambitions of your side gig could be a lot more grand.
It’s still early, of course. The bots themselves are expensive as hell if you’re running the latest versions of Claude Code for all of them. Getting this stuff running is hard; you’re bouncing between obscure references to Gas Town on Steve Yegge’s Github, and a bunch of smart posts on Simon Willison’s blog, and sifting through YouTube videos about Ralph Wiggum to see if they’re about the Simpsons or the software.
It’s gonna be like that for a while, a little bit of a mess. But that’s a lot better than Enterprise Certified Cloud AI Engineer, Level II, minimum 11 years LLM experience required. If history is any guide, the entire first wave of implementations will be discarded in favor of more elegant and/or powerful second versions, once we know what we actually want. Build one to throw away. I mean, that’s kind of the spirit of the whole codeless thing, isn’t it?
This could all still sputter out, too. Maybe it’s another fad. I don’t love seeing some of the folks working on codeless tools pivot into asking folks to buy memecoins to support their expensive coding bot habits. The Big AI companies are gonna try to kill it or co-opt it, because tools that reduce the switching cost between LLMs to zero must terrify them.
But for the first time in a long time, this thing feels a little different. It’s emerging organically from people who don’t work for trillion dollar companies. It’s starting out janky and broken and interesting, instead of shiny and polished in a soulless live stream featuring five dudes wearing vests. This is tech made for people who like making things, not tech made for people who are trying to appease financiers. It’s for inventors, not investors.
I truly, genuinely, don’t care if you call it “codeless”; it just needs a name that we can hang on it so people know wtf we’re talking about. I worked backwards from “what could we write on a whiteboard, and everyone would know what we were talking about?” If you point at the diagrams and say, “The legacy code is complicated, so we’re going to do that as usual, but the client apps and mobile are all new, so we could just do those codeless and see how it goes”, people would just sort of nod along and know what you meant, at least vaguely. If you’ve got a better name, have at it.
In the meantime, though, start hacking away. Make something more ambitious than you could do on your own. Sneak an army of bots into work. Build something that you would have needed funding for before, but don’t now. Build something that somebody has made a horrible proprietary version of, and release it for free. Share your Markdown files!
Maybe the distance from idea to app just got a little bit shorter? We're about to find out.
2026-01-15 08:00:00
When Wikipedia launched 25 years ago today, I heard about it almost immediately, because the Internet was small back then, and I thought “Well… good luck to those guys.” Because there had been online encyclopedias before Wikipedia, and anybody who really cared about this stuff would, of course, buy Microsoft Encarta on CD-ROM, right? I’d been fascinated by the technology of wikis for a good while at that point, but was still not convinced about whether they could be deployed at such a large scale.
So, once Wikipedia got a little bit of traction, and I met Jimmy Wales the next year, I remember telling him (with all the arrogance that only a dude that age can bring to such an obvious point) “well, the hard part is going to be getting all the people to contribute”. As you may be aware, Jimmy, and a broad worldwide community of volunteers, did pretty well with the hard part.
Wikipedia has, of course, become vital to the world’s information ecosystem. Which is why everyone needs to be aware of the fact that it is currently under existential threat from those who see any reliable source of truth as an attack on their power. The same authoritarians in power who are trying to purchase every media outlet and social network where ordinary people might have a chance to share accurate information about their crimes or human rights violations are deeply threatened about a platform that they can’t control and can’t own.
Perhaps the greatest compliment to Wikipedia at 25 years old is the fact that, if the fascists can’t buy it, then they’re going to try to kill it.
What I couldn’t foresee in the early days, when so many were desperate to make sure that Wikipedia wasn’t treated as a credible source — there were so many panicked conversations about how to keep kids from citing the site in their school papers — was how the site would become infrastructure for so much of the commercial internet.
The first hint was when Google introduced their “Knowledge Panel”, the little box of info next to their search results that tried to explain what you were looking for, without you even having to click through to a website. For Google, this had a huge economic value, because it kept you on their search results page where all their ad links lived. The vast majority of the Knowledge Panel content for many major topics was basically just Wikipedia content, summarized and wrapped up in a nice little box. Here was the most valuable company of the new era of the Internet, and one of their signature experiences relied on the strength of the Wikipedia community’s work.
This was, of course, complemented by the fact that Wikipedia would also organically show up right near the top of so many search results just based on the strength of the content that the community was cranking out at a remarkable pace. Though it probably made Google bristle a little bit that those damn Wikipedia pages didn’t have any Google ads on them, and didn’t have any of Google’s tracking code on them, so they couldn’t surveil what you do when you were clicking around on the site, making it impossible for them to spy on you and improve the targeting of their advertising to you.
The same pattern played out later for the other major platforms; Apple’s Siri and Amazon’s Alexa both default to using Wikipedia data to answer common questions. During the few years when Facebook pretended to care about misinformation, they would show summaries of Wikipedia information in the news feed to help users fact-check misinformation that was being shared.
Unsurprisingly, a lot of the time when the big companies would try to use Wikipedia as the water to put out the fires that they’d started, they didn’t even bother to let the organization know before they started doing so, burdening the non-profit with the cost and complexity of handling their millions of users and billions of requests, without sharing any of their trillions of dollars. (At least until there was public uproar over the practice.) Eventually, Wikimedia Foundation (the organization that runs Wikipedia) made a way for companies to make deals with them and actually support the community instead of just extracting from the community without compensation.
Things had reached a bit of equilibrium for a few years, even as the larger media ecosystem started to crumble, because the world could see after a few decades that Wikipedia had become a vital and valuable foundation to the global knowledge ecology. It’s almost impossible to imagine how the modern internet would function without it.
But as the global fascist movement has risen in recent years, one of their first priorities, as in all previous such movements, has been undermining any sources of truth that can challenge their control over information and public sentiment. In the U.S., this has manifested from the top-down with the richest tycoons in the country, including Elon Musk, stoking sentiment against Wikipedia with vague innuendo and baseless attacks against the site. This is also why Musk has funded the creation of alternatives like Grokipedia, designed to undermine the centrality and success of Wikipedia. From the bottom-up, there have been individual bad actors who have attempted to infiltrate the ranks of editors on the site, or worked to deface articles, often working slowly or across broad swaths of content in order to attempt to avoid detection.
All of this has been carefully coordinated; as noted in well-documented pieces like the Verge’s excellent coverage of the story, the attack on Wikipedia is a campaign that has been led by voices like Christopher Rufo, who helped devise campaigns like the concerted effort to demonize trans kids as a cultural scapegoat, and the intentional targeting of Ivy League presidents as part of the war on DEI. The undermining of Wikipedia hasn’t yet gotten the same traction, but they also haven’t yet put the same time and resources into the fight.
There’s been such a constant stream of vitriol directed at Wikipedia and its editors and leadership that, when I heard about a gunman storming the stage at the recent gathering of Wikipedia editors, I had assumed it was someone who had been incited by the baseless attacks from the extremists. (It turned out to have been someone who was disturbed on his own, which he said was tied to the editorial policies of the site.) But I would expect it’s only a matter of time until the attacks on Wikipedia’s staff and volunteers take on a far more serious tone much of the time — and it’s not as if this is an organization that has a massive security budget like the trillion-dollar tech companies.
The temperature keeps rising, and there isn’t yet sufficient awareness amongst good actors to protect the Wikipedia community and to guard its larger place in society.
Against this constant backdrop of increasing political escalation, there’s also been the astronomical ramp-up in demand for Wikipedia content from AI platforms. The very first source of data for many teams when training a new LLM system is Wikipedia, and the vast majority of the time, they gather that data not by paying to license the content, but by “scraping” it from the site — which uses both more technical resources and precludes the possibility of establishing any consensual paid relationship with the site.
A way to think about it is that, for the AI world, they’re music fans trading Wikipedia like it’s MP3s on Napster, and conveniently ignoring the fact there’s an Apple Music or Spotify offering a legitimate way to get that same data while supporting the artist. Hopefully the “Taylor’s Version” generation can see Wikipedia as being at least as worthy of supporting as a billionaire like Taylor Swift is.
But as people start going to their AI apps first, or chatting with bots instead of doing Google searches, they don’t see those Knowledge Panels anymore, and they don’t click through to Wikipedia anymore. At a surface level, this hurts traffic to the site, but at a deeper level, this hurts the flow of new contributors to the site. Interestingly, though I’ve been linking to critiques of Wikipedia on my site for at least twenty years, for most of the last few decades, my biggest criticism of Wikipedia has long been the lack of inclusion amongst its base of editorial volunteers. But this is, at least, a shortcoming that both the Wikimedia Foundation and the community itself readily acknowledge and have been working diligently on.
That lack of diversity in editors as a problem will pale in comparison to the challenge presented if people stop coming to the front door entirely because they’re too busy talking to their AI bots. They may not even know what parts of the answers they’re getting from AI are due to the bot having slurped up the content from Wikipedia. Worse, they’ll have been so used to constantly encountering hallucinations that the idea of joining a community that’s constantly trying to improve the accuracy of information will seem quaint, or even absurd, in a world where everything is wrong and made up all the time.
This means that it’s in the best interests of the AI platforms to not only pay to sustain Wikipedia and its community so that there’s a continuous source of new, accurate information over time, but that it’s also in their interest to keep teaching their community about the value of such a resource. The very fact that people are so desperate to chat with a bot shows how hungry they are for connection, and just imagine how excited they’d be to connect with the actual humans of the Wikipedia community!
It’s easy to forget how radical Wikipedia was at its start. For the majority of people on the Internet, Wikipedia is just something that’s been omnipresent right from the start. But, as someone who got to watch it rise, take it from me: this was a thing that lots of regular people built together. And it was explicitly done as a collaboration meant to show the spirit of what the Internet is really about.
Take a look at its history. Think about what it means that there is no advertising, and there never has been. It doesn’t track your activity. You can edit the site without even logging in. If you make an account, you don’t have to use your real name if you’d like to stay anonymous. When I wrote about being the creator of an entirely new page on Wikipedia, it felt like magic, and it still does! You can be the person that births something onto the Internet that feels like it becomes a permanent part of the historical record, and then others around the world will help make it better, forever.
The site is still amongst the most popular sites on the web, bigger than almost every commercial website or app that has ever existed. There’s never been a single ad promoting it. It has unlocked trillions of dollars in value for the business world, and unmeasurable educational value for multiple generations of children. Did you know that for many, many topics, you can change your language from English to Simple English and get an easier-to-understand version of an article that can often help explain a concept in much more approachable terms? Wikipedia has a travel guide! A dictionary! A collection of textbooks and cookbooks! Here are all the species! It’s unimaginably deep.
Whenever I worry about where the Internet is headed, I remember that this example of the collective generosity and goodness of people still exists. There are so many folks just working away, every day, to make something good and valuable for strangers out there, simply from the goodness of their hearts. They have no way of ever knowing who they’ve helped. But they believe in the simple power of doing a little bit of good using some of the most basic technologies of the internet. Twenty-five years later, all of the evidence has shown that they really have changed the world.
If you are able, today is a very good day to support the Wikimedia Foundation.
2026-01-12 08:00:00
Last week, we talked about one huge question, “How the hell are you supposed to have a career in tech in 2026?” That’s pretty specific to this current moment, but there are some timeless, more perennial questions I've been sharing with friends for years that I wanted to give to all of you. They're a short list of questions that help you judge whether a job that you’re considering is going to crush your soul or not.
Obviously, not everyone is going to get to work in an environment that has perfect answers to all of these questions; a lot of the time, we’re lucky just to get a place to work at all. But these questions are framed in this way to encourage us all to aspire towards roles that enable us to do our best work, to have the biggest impact, and to live according to our values.
This question originally started for me when I would talk to people about new startups, where people were judging the basic idea of the product or the company itself, but it actually applies to any institution, at any size. If the organization that you’re considering working for, or the team you’re considering joining, is able to achieve their stated goals, is it ultimately going to have a positive effect? Will you be proud of what it means? Will the people you love and care about respect you for making that choice, and will those with the least to gain feel like you’re the kind of person who cares about their impact on the world?
Where does the money in the organization really come from? You need to know this for a lot of reasons. First of all, you need to be sure that they know the answer. (You’d be surprised how often that’s not the case!) Even if they do know the answer, it may make you realize that those customers are not the people whose needs or wants you’d like to spend most of your waking hours catering to. This goes beyond the simple basics of the business model — it can be about whether they're profitable or not, and what the corporate ownership structure is like.
It’s also increasingly common for companies to mistake those who are investing in a company with those who are their customers. But there’s a world of difference between those who are paying you, and those who you have to pay back tenfold. Or thousandfold.
The same goes for nonprofits — do you know who has to stay happy and smiling in order for the institution to stay stable and successful? If you know those answers, you'll be far more confident about the motivations and incentives that will drive key decisions within the organization.
Now we’re getting a little bit deeper into thinking about the systems that surround the organization that you’re evaluating. Every company, every institution, even every small team, is built around a set of invisible assumptions. Many times, they’re completely reasonable assumptions that are unlikely to change in the future. But sometimes, the world you’re working in is about to shift in a big way, or things are built on a foundation that’s speculative or even unrealistic.
Maybe they're assuming there aren't going to be any big new competitors. Perhaps they think they'll always remain the most popular product in their category. Or their assumptions could be about the stability of the rule of law, or a lack of corruption — more fundamental assumptions that they've never seen challenged in their lifetime or in their culture, but that turn out to be far more fragile than they'd imagined.
Thinking through the context that everyone is sharing, and reflecting on whether they’re really planning for any potential disruptions, is an essential part of judging the psychological health of an organization. It’s the equivalent of a person having self-awareness, and it’s just as much of a red flag if it’s missing.
Here is how we can tell the culture and character of an organization. If you’ve got connections into the company, or a backchannel to workers there, finding out as much information as you can about the real story of its working conditions is often one of the best ways of understanding whether it’s a fit for your needs. Now, people can always have a bad day, but overall, workers are usually very good at providing helpful perspectives about their context.
And more broadly, if people can provide examples of those in power within an organization using that power to take care of their workers or customers, or to fight for the company to be more responsible, then you’ve got an extremely positive sign about the health of the place even before you’ve joined. It’s vital that these be stories you are able to find and discover on your own, not the ones amplified by the institution itself for PR purposes.
And here we have perhaps one of the easiest and most obvious ways to judge the culture of an organization. This is even a question you can ask people while you’re in an interview process, and you can judge their responses to help form your opinion. A company, and leadership culture, that can change its mind when faced with new information and new circumstances is much more likely to adapt to challenges in a healthy way. (If you want to be nice, phrase it as "What is a way in which the company has evolved or changed?")
This is where we go from the abstract and psychological goals to the practical and everyday concerns: can you pay your bills? The phrasing and framing here is very intentional: are they really going to pay you enough? I ask this question very specifically because you’d be surprised how often companies actually dance around this question, or how often we trick ourselves into hearing what we want to hear as the answer to this question when we’re in the exciting (or stressful) process of considering a new job, instead of looking at the facts of what’s actually written in black-and-white on an offer letter.
It's also important not to get distracted with potential, even if you're optimistic about the future. Don’t listen to promises about what might happen, or descriptions of what’s possible if you advance in your role. Think about what your real life will be like, after taxes, if you take the job that they’ve described.
This is where you can apply your optimism in a practical way: can the organization accurately describe how your career will proceed within the company? Does it have a specific and defined trajectory, or does it involve ambiguous processes or changes in teams or departments? Would you have to lobby for the support of leaders from other parts of the organization? Would making progress require acquiring new skills or knowledge? Have they committed to providing you with the investment and resources required to learn those skills?
These questions are essential to understand, because lacking these answers can lead to an ugly later realization that even an initially-exciting position may turn out to be a dead-end job over time.
Sometimes it can really feel like the deck is stacked against you when you're trying to find a new job. It can feel even worse to be faced with an opportunity and have a nagging sense that something is not quite right. Much of the time, that feeling comes from the vague worry that we're taking a job that is going to make us miserable.
Even in a tough job market, there are some places that are trying to do their best to treat people decently. In larger organizations, there are often pockets of relative sanity, led by good leaders, who are trying to do the right thing. It can be a massive improvement in quality of life if you can find these places and use them as foundations for the next stage of your career.
The best way to navigate towards these better opportunities is to be systematic when evaluating all of your options, and to hold out for as high standards as possible when you're out there looking. These seven questions give you the tools to do exactly that.