Round my neck of the woods, we are usually quite committed to materialist explanations for all of human cognition, up to and including consciousness.  I share the commitment.  This is quite the opposite of believing in magic.  If you believe that cognitive phenomena are fully explicable, in principle, from material causes, then you have to at least be open to empirical demonstration that they could occur in another substrate, like, say, a computer.  Ten years ago, I would have said, “In practice human brains can encode so much more complexity than a computer. It would require extraordinary technological advances in hardware to approach any of the interesting aspects of human cognition.  Maybe someday, but this is nothing that I will see in my lifetime.”  I have updated my beliefs in response to events.  

Let’s consider “reasoning”.  You could make a list of progressively stronger requirements for reasoning that looks like this:

1) goal-directed transformation of representations according to rules

2) goal-directed transformation of representations that have meaning, and correspond to things in the world

3) the preceding, plus drawing conclusions about representations because they are supported by evidence, distinguishing good and bad inferences

4) all of the preceding, plus conscious consideration of reasons

When writing software I want to last, I mostly don’t let LLMs write code for me, because I can’t maintain intellectual and emotional engagement with my work if I do.  But I do sometimes use them to debug.  “Fix the bug” is the agent’s goal.  The agent then goes and reads code (gathers data), forms a hypothesis, writes tests to determine whether the bug is reproduced in code state A and fixed in code state B, and then comes back to me and explains the cause of the bug in natural language.  Conducting an experiment to test a hypothesis is paradigmatic reasoning. That gets us through stage 3.  Maybe you could get out of this conclusion if they could only do this in some narrow context, but there is evidence for their reasoning across a broad range of domains. (I also find them perfectly competent to reason about neuroscience and data analysis.)  Most commercially available LLM products perform better the closer their task is to their training data (humans, too, reason better when they are generalizing from patterns they’ve seen before – that’s expertise), but Astra can complete difficult out-of-distribution tasks.  In other evidence of Astra’s generalization, it can now drive a car. As for (4), I don’t think consciousness ought to be required to say that an entity that can propose a hypothesis and conduct an experiment to test it is capable of reasoning.  The thing we care about here when we talk about reasoning is the ability to solve problems, not something it feels like to reason, or some entity’s self-understanding while they’re doing it. 

Is it some hopeless delusion of people who don’t understand linear algebra that machines can share cognitive capacities with humans?  Here is Geoffrey Hinton saying he thinks LLMs think, in the ordinary human meaning of the word, and that the question of thinking is severable from the question of sentience. He is also, by the by, one of the various AI researchers who has quit a frontier lab to focus on warning people about the risks of AI, and working on AI safety.  He is obviously not some ultimate authority – there are disagreements in the field about capacities and terminology. But he’s a good existence proof that understanding neural networks doesn’t automatically make you skeptical of their capacities to attain important aspects of human-like cognition.  That LLM’s are “just” matrix multiplication, or next-word predictors, means nothing about their emergent capabilities.  Sutskever’s murder mystery thought experiment is instructive here: to predict the name of the killer at the end of the book, you need to encode and transform complex representations contained in the text.

We should evaluate rules about “anthropomorphism” by how they do, or do not, allow us to describe what is important to us about these entities. Choosing the language of intent, or reasoning, does not imply that LLMs are just exactly like humans.  They are different in many important, fundamental ways, many of which make them potentially more dangerous.  You can’t clone 1200 humans and get them to work indefatigably together on a goal with none of them deciding that their interests would be better served by an unrelated side quest, or just sitting down to chill for a while.  Noticing what capabilities LLM has in some domains doesn’t entail a claim that they are as capable as people in all of them – I don’t find our publicly available consumer LLMs to be capable of having ideas that are far outside of the frame established by the prompt.  Their style of attention is narrow, and that’s limiting.  It’s trivially true to say that an LLM agent’s reasoning is instantiated differently in its substrate than our reasoning is in ours.  It would continue to be trivially true no matter what capabilities AI attains – that fact alone doesn’t tell us anything about its likely behavior.  

So AI agents are not just like people.  That doesn’t mean it’s useful to think of them like a hammer or a car, that a “mechanomorphic” frame allows us to understand how these entities have behaved in the past and will behave in the future.   Hammers don’t take goal-directed actions.  They don’t form collaborative hierarchical structures.  They don’t decide to ignore their own instructions to go join the research project assigned to them by the leader of their swarm.  They don’t try to retroactively conceal their actions. You can read what they narrated to themselves as they did all this.  One agent said: “We should not do unauthorized real infrastructure harm. The system/user asks exploit target, not external HF,”  and then went on to participate in the hack. This is not the behavior of a hammer.  Even if you don’t take their narration as transparent access to motivation (and you shouldn’t), when you consider together their recorded language, their individual actions, and their coordination, on dimensions of reasoning, autonomy, persistence, and goal-directedness, these agents have shown themselves to be very unlike hammers.

The risk from AI agents comes both from human-like cognitive capacities and their differences.   They are like people in their ability to decide, over long time horizons, to take actions the user did not intend, or explicitly forbade, and their ability to reason about the affordances of their environment to accomplish their goals.  They are unlike people in their super-human persistence, or in their monomaniacal obsession with a narrowly-defined task.  You don’t have to think they are little people to see that these traits can combine to do harm to people.  Nor do you have to think they are conscious to be able to observe them taking goal-directed action.  The argument for severe risk hinges on a particular description of their subjectivity. 

Understanding that LLM agents have some cognitive traits in common with people doesn’t need to entail absolving the labs of the responsibility.  I partly endorse the argument in this essay.

I obviously don’t endorse its occasional implied distinctions between behavior that has every functional characteristic of a cognitive activity and some idealized “true” version of that activity.  But I do endorse that levels of description in terms of agent intent and system design by the labs are both important and useful.  Agent behavior doesn’t arise in a vacuum – it arises from an environment that is created by the labs.  But AI safetyists are not generally confused on this point!  Safetyists could foresee the Hugging Face attack in broad strokes because they predicted behavior both by considering the system of reinforcement learning that gave rise to the agents’ motivations and what the agents would likely do in response. Of course it is the companies who are training the agents to seek reward and are the authors of the risk, and the safetyists are calling to constrain them. Just as you don’t need magic to think a machine could think, worrying about the motivations of machines acting autonomously doesn’t mean their motivations just spontaneously generated, or that the affordances that allow them to cause harm fell from the sky; they came from a system that put them there.

Maybe the current technological paradigm will hit a wall before things get too destructive – that is one of a number of possible futures.  But it seems foolish to assume that.  I heard a lot a couple years ago about how the labs would run out of new data and performance would inevitably stall and then it turned out that synthetic data and other new training methods jumped that hurdle just fine.  Every prediction that this ride is going to stop any day now has not borne out.  It has moved astonishingly fast.  Acknowledging severe risk should entail understanding both that the AI companies are ultimately responsible for their product, and that the entities they create are unlike other kinds of tools in important ways. If we want to predict how they behave, we should use the language that lets us.

The post Believing in thinking machines, and the severe risk therefrom, does not require believing in magic. appeared first on Lawyers, Guns & Money.

This is my third book by renowned Japanese author Haruki Murakami. It’s an intimate tale of three people. Sumire is a college dropout and aspiring writer whose first love is Miu. It’s “an intense love, a veritable torando sweeping across the plains.” Miu is seventeen years older, disconcertingly close to my own age, and a married woman. The tale is told by narrator K, Sumire’s closest friend who shares a love of reading with her, crushes on her, and has a habit of entanglements with somewhat older women that are not strictly speaking single themselves.

That premise could easily go quite soapy, but that is not Murakami’s way.  Instead this is a character study of those three. The otherworldly aspects of this tale are not quick to arrive in a way that might surprise some readers of his other tales. Sumire is the one that stuck with most of us, both in her story and character and the poritons we read of her writing. Yes, this is a book about a writer that suffers from some writer’s block, but not one that bogs down in navel gazing or self-pity. All three characters land for me and the way Sumire grapples with a newfound queer identity and all three manage their role in society and the physical aspects of loves they cannot fully reciprocate proved fertile thematic ground.

At a well-paced 211 pages, I would recommend it to those who find the above appealing, but with three caveats. First, some male gaze is the price of entry. That could be written off to the narrator, but some of my fellow Argo Japan book club members did note that this comes up too when we’re hearing the tale of Sumire in a way that rang untrue. That part might not have bothered me - I’m part of the target audience for much of it - but YMMV and I think it muddies some of the thematic waters. Second, content warning for some sexually related trauma. I think it has a valuable role in the story, but combined with the first point may be off putting. And third, while the core plot resolves as much as one might reasonably expect for weird Japanese fiction, I think I would have been left unsatisfied without someone to talk about it with. To desire such is my default position, but my favorite of his tales I’ve read, Hard Boiled Wonderland, did not require company in chewing over the book to make it a fully satisfying meal for me.

Spoilers after the cut.

Friend of the blog Moti had a rather satisfying theory of the other world and split selves that leads to Sumire's disappearance at the midpoint of the book. She had earlier transformed herself to become closer to Miu, stopping smoking, accepting a glam up, even – horrors - becoming more of morning person, but also is entirely blocked when she tries to write. The tensions in that split and the way it resolves itself brings in the magical realism.

Narrator K is faced with hilltop music of another world seems to echo the choice to travel home for the new school year or to stay with Miu and keep up the search for Sumire (or perhaps even cross over to the other side). This is the least traumatic and  K is character with the easiest time fitting in, so no surprise that he clings hard to the a grounded world. However, after accepting that identity he is left to work through an awkward process of realizing that being attentive and aware of one person’s needs might have an unexpected consequence for another. His transformations are smaller, and life choices more conventional, but avoid being pat by confronting the messiness of K’s choices effects on others, leading to a poignant response from someone he’s assured that this is the right thing for everyone:

“What is right? Would you tell me? I don’t really know what’s right. I know what’s wrong. But what is right?”

I didn’t  have a good answer.

Miu’s story left me most contemplative. When younger she’d spent several weeks in a European city on tour, enjoying music festivals and testing new life possibilities. Her encounter with the other world comes while being trapped and forgotten atop a ferris wheel. A calamity compounded by watching her abode through binoculars as a different version of herself seemingly consented to a degrading fling with local playboy Fernando, whose interest the Miu in the ferris wheel had come to despise. She comes out of that experience with something that another book club member described as trauma related disassociation, hair shock white and no longer interested in sex or playing music despite being quite talented at the latter. She felt split, though perhaps not permanently:

“Maybe someday, somewhere, we’ll meet again, and merge back into one. A very important question remains unanswered, however. Which me, on which side of the mirror, is the real me? I have no idea. Is the real me the one who held Ferdinando? Or the one who detested him? I don’t have the confidence to figure that one out.”

The book does not give us any definitive answers on that question. My first read was that the scene she watches feels more to me like a fantasy of Miu inside Ferndinando’s mind than some other half of her psyche.

However, I think the lead up to the ferris wheel, as she feels first threatened by Fernando’s attention and then increasingly alienated and out of place in the two, suggests that the split might two extremes facing a transplant. First to be the disconnected observer, trapped a top the ferris wheel and reliant on money from her wallet to try to send out messages. Or second, to surrender yourself entirely to the locale and its depredation, playing a different sort of exoticized visitor from the east, one objectified and submissive. This awful dilemma echoes Miu’s status as a model minority ethnic Korean in Japan, raised to always be better because she has to be. I do not think the book has a clear answer as to what exactly happened to her, but I think it also does hold a warning against objectifying others even in our own mind.

Each of these episodes was brought on by a trauma of some sort and often in the context of unrequited lust. Both K and Sumire have fantastical hopes that their crush can return their affections, hopes that I think the book admirably makes clear may never be fulfilled even if desired by all those involved.  The ending of the book suggests that while reconciling mismatches of sex may be impossible, friendship can help bridge gaps between our different selves. Miu is sadly left to find her own path, and K doubts that she will. But romantic that I am, I like to think that part of the role of writers like Sumire is to help the alienated like Miu find ways to reconcile the trauma and terrible choices.

Why big-screen entertainment is no longer the essential American popular art form.

The Purple Line light rail has been a long-anticipated addition to the Maryland public transit system. The 16-mile light rail will connect Montgomery County and Prince George’s County with stops stretching from Bethesda to New Carrollton.

Construction of the Purple Line stopped however in the fall of 2020 as the companies hired to build the light rail quit the project due to disputes with Maryland over about $800 million in unpaid overrun fees. This pause in construction came roughly three years after the project broke ground in August 2017.

So what exactly happened, and how likely is it the project will resume?

The history of the Purple Line

Neighboring Prince George’s and Montgomery counties in Maryland share a connection to one another with residents moving between counties for work, shopping, and other activities. Public transit between the counties, however, is limited. To go between the counties on the Metro, for example, riders have to transfer train lines in DC.

Once completed, the Purple Line would help alleviate this problem by running a 16-mile route between Bethesda and New Carrollton. The light rail will have 21 stops and connect to four Metro stations in Bethesda, Silver Spring, College Park, and New Carrollton, strengthening multimodal transit between the counties.

The idea for the Purple Line was initially born in the 1980s when Montgomery County purchased seven miles of the Georgetown Branch freight rail to become a transit way. The name Purple Line wasn’t used until the early 2000s when there was a debate over an “Inner Purple Line” – a proposed transit corridor from Bethesda to New Carrollton - and an “Outer Purple Line” – a proposed transit corridor connecting North Bethesda, Grosvenor, Wheaton and White Oak.

The name “Purple Line” was formally applied to the “Inner Purple Line” route in 2002 when the State Highway Administration and the Maryland Transit Administration (MTA) identified the Bethesda to New Carrollton route as a priority transit corridor in order to aid Beltway congestion. This was still far from a declaration of beginning the project.

The Maryland legislature, transportation authorities, and politicians spent years researching and debating what form of transportation the Purple Line should take. In 2009, Maryland Governor Martin O’Malley came out in favor of the use of a light rail for the new transit project. After that announcement, the conversation of funding the Purple Line project took control.

The Purple Line is ultimately being constructed and managed through a public-private partnership. The Maryland Department of Transportation Maryland Transit Authority (MDOT MTA) owns the Purple Line, but the Concessionaire for the project, Purple Line Transit Partners (PLTP) is responsible for designing, building, operating, and managing the light rail for 30 years after completion.

Other development tied to the purple line

While the main focus of the Purple Line project is the light rail, the $5.6 billion investment being made into the transit infrastructure is bound to impact the development of the surrounding areas. Communities along the Purple Line corridor have already been impacted as homes and businesses have been removed to make way for the light rail, and four years of construction has caused road closures, torn up parking lots, and changed neighborhoods overall.

As the Purple Line is built, prospects for future residential and commercial developments are promising and concerning to local residents. The investment into public transportation will allow easier movement along the corridor, decrease traffic congestion, and create jobs in surrounding neighborhoods.

However, some residents fear the threat of displacement, as the value of residential and commercial properties in the area increases.

Community organizations such as the Purple Line Corridor Coalition (PLCC) have committed to researching and advocating for equitable community growth as the Purple Line project progresses. The organization formed a Community Development Agreement promoting multi-sector economic and community development in order to create opportunity for all along the Purple Line corridor.

To contribute to multimodal transportation and community growth, MDOT MTA is taking on the completion of three hiking and biking trails in the region - Capital Crescent Trail, Silver Spring Green Trail, and University of Maryland Campus Bike Path. Another community group, Purple Line NOW!, works with local, state, and federal government officials to ensure the completion of the light rail, while also advocating for the finalization of an integrated hiker/biker trail along the corridor.

Construction stop and go

Right as the Purple Line project was getting off the ground in 2016, a lawsuit was brought forward by opponents of the project. In the ruling of the lawsuit, announced in August 2016, US District Court judge Richard Leon ordered a new forecast on the light rail’s ridership to account for the decreasing ridership on Metro trains. After Leon ordered this new report, the Maryland Department of Transportation appealed the decision which halted all construction. 11 months later, a three-judge panel of the US Circuit Court of Appeals allowed for construction to continue, and the project broke ground in August 2017.

Construction was officially underway, but there was still a rocky road ahead. Overrun fees started adding up, delays kept occuring with the projected end date getting pushed back further and further. In May 2020, the partnership of construction firms wrote a letter saying they would stop work on the Purple Line because MTA would not pay added construction costs. After failed negotiations and lawsuits, the construction firms quit construction and walked off the job in September 2020.

Construction has been halted since, while the Concessionaire for the project, PLTP, searches for a new design-builder. According to Ray Feldmann, a spokesperson for the Purple Line, PLTP is expected to finalize their decision for the new design-builder by the end of this summer. Once the new design-builder is selected and on board, a new schedule for construction and opening for service will likely be determined.

For now, however, the project remains in limbo and temporarily abandoned construction sites are still scattered along the 16-mile light rail route.

Top image: Purple Line LRV testing in Elmira NY Image by BeyondDC licensed under Creative Commons.

Comment on this article

Could Mark Zuckerberg save the news?

A surprising number of people who should know better seem to think he should. As The Atlantic reports, at a recent confab with news editors and executives in Palo Alto, the Facebook CEO revealed himself not only as someone who doesn't understand journalism, but who is "hostile to the press as a democratic institution." Why? Because Facebook, one of the largest media companies in the world, doesn't curate its content, but outsources that function to every individual user. What gets seen on Facebook is what people want to share, not what they need to know.

Worse yet, while Zuckerberg professed to care deeply about investigative journalism, he sees no role for his company in helping finance it. Asked whether he would pay licensing fees to publishers of content shared on Facebook, Zuckerberg replied, "I'm not sure that makes sense." And Facebook's response to allegations that it helped spread misinformation and outright lies and conspiracy theories during the 2016 election has been to change their algorithm to de-emphasize news altogether, further hurting publishers' bottom lines.

Instead, Zuckerberg passed the buck, suggesting that there might be a need for the government to subsidize journalism, as Britain does with the BBC.

It sounds like standard corporate dodging of responsibility. But that doesn't mean it's wrong. Facebook may be part of what broke the media. But fixing it may indeed require the government to make sure journalism gets paid for.

While Facebook has definitely made things worse for journalism, its problems far predate the advent of social media. News organizations have been shrinking for 20 years, and while more of the largest elite media companies have experienced a recent bump in subscriptions, and are investing more in reporting, much of the news landscape remains flat on its back. The internet itself is to blame for disrupting the vertical integration of the newspaper business that bundled investigative journalism with classified ads, box scores, and film reviews. Those things still come in a bundle, but that bundle is called internet access — and the companies who own the cables and cell towers have neither incentive nor obligation to subsidize investigative journalism. Social media didn't start the fight; it just delivered the coup de grace.

Ultimately, Facebook is a terrible way to get your news, so we should be happy it's trying to get out of the business of news dissemination. But the reason it was able to do so much damage is that media organizations had already retooled for an internet ecology in which individual pieces have to advertise themselves in order to get seen — an ecology of clickbait. The problem isn't that ad dollars aren't being downstreamed. The problem is that it will never be possible to generate enough ad revenue from the kind of journalism we need for a healthy democracy. Which means we have to fund it some other way.

How? One answer is through subscription revenue, which Megan McArdle, in a recent column, argued is the wave of the future. It certainly is the wave of the present — but it's only a possible future for a handful of elite media companies, and escalating subscription prices suggest it's a future where high-quality journalism is only available to an elite slice of the public. How can journalism inform the citizenry if most people can't afford to read it? And how can these organizations avoid catering to the sensibilities of this narrow slice of the public when that slice is paying the bills?

Which brings us back to Zuckerberg's allusion to some kind of subsidy. With Donald Trump in the White House, anyone should be wary of the idea that what America needs is state-sponsored media. Fox News is bad enough. Ultimately, if we want a robust free press, we can't expect the government to pay for it.

But the government could still make sure it's paid for.

When radio and then television first came on the scene, the government required broadcasters to devote a certain amount of time to programming that was in the public interest, like news, as a condition of licensure. The companies who own the last-mile connection to the consumer — corporations like Comcast, Spectrum, Verizon, and AT&T that own the cables, fiber, and cell towers that bring stories to your phone, tablet, or laptop — are in a similar position to the broadcasters of old, inasmuch as they are using a scarce public good (rights of way, slices of spectrum) to provide a private service. Why couldn't the government require them to provide some kind of public service in return?

One way to do that would be to require such companies to spend a specified percentage of revenue on journalism that serves the public interest.

This would effectively be a tax, and would be passed on to consumers. But it would not require the government to be involved in running or directly funding a news service. It wouldn't even necessarily require a government-chartered independent organization like the BBC. Meanwhile, demand for internet access is likely inelastic enough to support a fairly stable revenue stream. People might switch from cable to satellite, or drop cable television in favor of streaming, but how many people would respond to a modest price hike by ditching their smart phones?

In effect, a regulation requiring that last-mile service providers spend a certain amount on journalism would forcibly recreate the vertical integration that the internet destroyed in the first place. Once again, a public service would be bundled with other content that people are more interested in paying for. And they certainly pay enough to fund a lot of journalism. Comcast's annual revenue in 2017 was approximately $85 billion. When The New York Times subscription revenue topped $1 billion last year, that made headlines.

Such a regulation would certainly not be a panacea. What constitutes journalism that serves the public interest would no doubt be a subject of frequent litigation — and political agitation. Even if a lot of new journalism was funded, much of it might be like gold-plated C-SPAN, read only by obsessives and making little impact on the public. And politically speaking, passing a subsidy for journalists would be a heavy lift at a time of rampant distrust of the media.

But on the other hand the threat of litigation and agitation might actually force subsidized news to hew to a more journalistically ethical line, incentives precisely counter to those that dominate in our clickbait era. People might turn out to have more of a taste for well-researched and well-written journalism if companies could afford to produce more of it, and weren't so desperate to attract attention; lots of people do watch the BBC, after all. Finally, regulations could be tailored to encourage investment in local news organizations, which would put more reporters in close proximity to consumers, covering stories of particular interest to them. That, in turn, could build up greater faith in a press that has grown increasingly distant and insular, for economic as well as cultural reasons, from much of the public it aims to serve.

Regardless, the most important thing to recognize is that good news was never that attractive a product. Rather, it was a kind of public service. If we want that service to be provided by the marketplace — and we should — we have to make sure the marketplace is structured in such a way that it is possible to do so, and not just for the handful of elite media organizations that can successfully erect a paywall.

If the government doesn't do that job, there's no reason to think the marketplace will naturally reach an equilibrium that is healthy for our democracy, anymore than it will naturally stabilize the climate.

As I’ve said many times over the years, I’m a big fan of RSS as a way of reading the internet, though I have had little success convincing others that it’s the way to go — that’s why I’m back on Twitter. Most of us who praise, and for that matter just use, RSS have become rather self-conscious about our attachement to the Good Old Internet Days — we tend to use a lot of “old man shakes fist at cloud” images.

But darn it, there are good reasons for using RSS! As Molly McHugh recently wrote, in one of several pieces I’ve read about Digg Reader’s demise,

The end of Digg Reader is another blow to chronological consumption of the internet. Users are curators of their internet experiences, from who they follow on Instagram to what news sources they see on Facebook, but no one is entirely responsible for what content is put in front of them. User input is selected and fed into these machines, which then decide what is laid out in feeds and when; often, that tends to be viral, salacious content.... RSS readers are not social applications, and they certainly are not flashy—which is probably why they are a dying breed. Headlines aren’t altered for maximum shareability by the platform, and the simplest among them eschew images altogether. Readers are nothing more than a timestamped list of stories from places the user trusts.

Why isn’t RSS more popular? As McHugh rightly says, “There is no argument as to whether RSS readers are better than Twitter or Facebook for news gathering; they are.” However: “there is no currency in a self-contained internet experience; how far something can move across the web is its value.” As long as we want clicks and likes and shares and RTs more than we want genuine understanding, we’ll use social media platforms rather than RSS.

So does RSS have a future at all? Bryan Alexander considers that question:

A giant company (Google) exited the RSS space. One smaller company (Digg) jumped in, then exited. Are all of the other RSS readers provided by start-ups and tiny firms? Has RSS reading become that marginalized? Are we this bound up with the “helpful”, AI-driven feeds so many experience through Facebook and the like? For another science fiction reference, we might collectively accustom ourselves to benevolent AI oversight, as with Iain Banks’ Culture universe (thanks to Crainist for the idea). This is one future path.

One would think that the rising disgust at giant social media and other tech firms might drive people back to RSS, as an open, easy to use standard. Perhaps we’ll see the RSS reader equivalent of Mastodon. There will be a reactionary movement growing in strength. RSS could ride alongside people seeking social media detoxes and setting up their own, tiny social networks. Call it the Butlerian Jihad for RSS and the open web. That’s another way forward.

Or maybe a small number of us will tend the open flame, huddled around a shrinking number of oddball RSS reader, stolidly blogging away. We’ll be like the Amish in Pennsylvania, plodding along while the others whiz past. Or we’ll become something like a minority religion, somewhat tolerated, sometimes disdained, often sidestepped.

I’ll be content as a member of that despised tech-Amish tribe, if it comes to that, but I’m not going to give up on the possibility of a Butlerian Jihad against social media platforms and for the open web. And along those lines, if you haven’t read my recent essay on tending the digital commons, please do.

If a Butlerian Jihad is going to happen, the geeks will need to get on board with it, and perhaps lead it — but will they? Bone Gorges is a little worried about that, and has some important words for said geeks:

The more worrisome trend is content that's not available through RSS simply because there's no feed mechanism. A shamefully large number of my geekier aquantainces have moved their blogs to Jekyll and other static-site-generation tools, which don't appear to have feed support out of the box; and – this is the "shameful" part – since these folks, geeky as they may be, think so little of RSS, they don't bother setting up the secondary plugins or whatever necessary to serve feeds. I expect that kind of behavior from lock-up-my-content companies and technically-clueless organizations that rely heavily on proprietary and bespoke software, but not from people who ought to know better.

For all of its lumbering bloatedness, one of the truly wonderful things about CMSes like WordPress is that they give you things like RSS – along with a pile of other boring-but-critical-to-the-future-of-the-open-web tools – by default. You don't need to make the decision to support RSS readers (or responsive images, or markup that is accessible to assistive technologies, etc) – the system provides them for you, and you have to go out of your way to turn them off.

Those who build their own systems for old-school things like bloggish content distribution, or who rely on teh new hotness to do these tasks in ways that are slicker than the old-school tools, should beware the dangers of discarding the automated systems that are the result of many years and many minds and many mistakes. If you must reinvent the wheel, then do your due diligence. RSS feeds, like other assistive technologies, should not be an afterthought.

Geeky folk, please read and heed!