Imagine coming across, on a reasonably serious site, an article that starts along the lines of:

After observing the generative AI space for a while, I feel I have to ask: does ChatGPT (and other LLM-based chatbots)… actually gablergh? And if I am honest with myself, I cannot but conclude that it sure does seem so, to some extent!

I know this sounds sensationalist. It does undermine some of our strongly held assumptions and beliefs about what does “to gablergh” actually mean — and what classes of entities can, in fact, be said to gablergh at all. Since gablerghing is such a crucial part of what many feel it means to be human, this is also certainly going to ruffle some feathers!

But here’s the thing: so far, after thousands of years of philosophical thought and scientific research, we have not been able to clearly define “gablerghing”. Thus, we simply cannot say for certain that some simpler animals, like ants, do not gablergh in some relevant sense. Gablerghing happens on a spectrum, from clearly gablerghing organisms like humans and dolphins, through animals like dogs or cats who I think we would mostly agree do gablergh, down to ants where this is maybe more fraught a statement.

So why couldn’t “a set of scripts running on top of a corpus of statistically analyzed internet content” be said to, in some sense, gablergh?

Naturally, your immediate reaction would not be to make a serious thinking face and consider deeply whether or not GPT indeed “gablerghs”, and if so to what degree. Instead, you would first expect the author to define the term “gablergh” and provide some relevant criteria for establishing whether or not something “gablerghs”.

Yet somehow when hype-peddlers claim that LLMs (and tools built around them, like ChatGPT) “think”, nobody demands of them clarification of what they actually mean by that, and what criteria they might possibly use (beyond “the output seems human-made”). This allows them to weaponize the complexity of defining the term “to think”, with all its emotional and philosophical baggage, and using it to their advantage.

“Well you can’t say it doesn’t think” — the argument goes — “since it’s so hard to define and delineate! Even ants can be said to think in some sense!”

This is preposterous. Of course this does not in any way prove that GPT can “think”; as one person pointed out on fedi it’s a case of the motte-and-bailey fallacy. Instead of accepting the premise, we should fire right back: “you don’t get to claim that GPT ‘thinks’ unless you first define that term clearly, and provide relevant criteria”. And these criteria need to be substantially better than the quack-like-a-duck of “output seems human-like, also it told me it thinks”.

After all, “at the very least you need to be able to define a quality Q you claim X has” is a much stronger stance than “I claim X has quality Q and you can’t prove I am wrong because Q is hard to define.”

No idea why we all collectively keep getting tripped over this, and fail to recognize it for what it is — thinly veiled hype-generation attempt that uses badly defined terms for marketing.

In the end, what it means to “think”, to be “conscious”, to “have intentionality”, is a matter for philosophers. Not for AI-techbros with stock to pump and chatbots to sell.

I remember trying to buy a TV that does not have “smart” functionality a few years ago. It was a chore. Today it seems nigh-impossible.

By the way, we need a nice way of referring to non-smart devices. I propose: “safe”.

And not just TVs: ovens; refrigerators; dishwashers — all are now “smart”. In fact, it seems that more and more the available non-smart, err, I mean safe models are only the simpler ones, less performant in ways that are not related to any smart functionality.

Safe TVs but without the fancy backlight. Safe refrigerators but without the de-icing system. My Safe TV was available only with lower resolutions than “smart” models of the same brand.

This really annoys me. I am too well aware of security implications of smart devices. I do not want to have to manage regular software updates for whatever number of appliances I have at home, or risk somebody using them in a botnet (or worse).

And no, I don’t trust their “disable WiFi” menu options either. Seen this setting get enabled without my consent too many times. And a lot of participants to my little completely unscientific fedi poll seem to have similar experiences. Plus, there is valid concern that some devices will just try to connect to any open WiFi network; I would much rather not

I could put such devices on a special VLAN, or behind a Pi Hole, but 99% of people can’t. Plus, it’s work. Plus, most importantly, you can bet that “smart” devices will start coming with SIM cards and 4g/5g modems very soon — cars already do. Why does my fridge need Internet connectivity in the first place?

In 2016 an IoT-based Mirai botnet took down Dyn, one of the biggest online infrastructure companies, and many well known websites with it.

As early as 2018 there were already botnets that… used CCTV cameras. But of course the predominant media narrative was “hackers attack” instead of “vendors put us at risk.”

Sidenote: if you’re using the word “hacker” to mean “cybercriminal”, you are making it worse. Please stop.

With all this in mind, I started thinking of how could this be solved? Not in the sense of “how can I, a techy person, secure my network and devices”, but in the sense of “how can we as a society manage the Internet of Shit problem?”

Consider a regulatory requirement for IoT / smart-appliance vendors to provide either (vendor’s choice):

• similarly-priced safe models, physically without the smart functionality, but with other metrics and functionality on-par with the smart version; or…
• reliable, verifiable, physical way of disabling smart functionality (or perhaps just networking) in their smart-devices.

Additionally, the packaging or other forms of information available before purchase should state clearly:

• does the device require Internet connectivity to set-up?
• does the device require a mobile app to set-up?
• does the device require agreeing to an EULA/TOS/privacy policy to set-up?
• which functions require Internet connectivity?
• which functions require data processing on external servers (that is, outside of the device)?
• does the device have a microphone or a camera or other sensors?
• does information from such sensors ever leave the device (for example, voice command data to be processed on external servers)?

I just want to be able to buy a damn refrigerator without worrying about it joining a botnet. Is that too much to ask?

This blogpost started off as a fedi thread. It got a bunch of interesting responses, and links to news about absolutely bonkers IoT stuff galore. Might be worth checking it out!

This week in the United States of America, a former British colony on the North American continent, long-brewing political and social problems culminated in a messy speaker election in the lower chamber of the bicameral national parliament.

The Republican party, by far the more conservative of the two major parties in what effectively is a two-party political oligopoly, gained narrow majority in the chamber in November elections, but was unable to effectively execute on its new-found power. A small far-right splinter group within the party blocked the election of the speaker — a procedural position that has gradually become heavily politicized — demanding political favors in return for their votes. This resulted in four days of heated and often chaotic proceedings, at one point devolving into a brawl.

The Speaker of the House, as the post is officially called, has finally been elected on the fifteenth try, the largest number since before the country’s bloody civil war. Last time election of the speaker — which is largely a formality — required more than one ballot was in 1923.

To placate the hold-outs, the now-newly elected speaker had to first agree to a long list of concessions, potentially going as far as giving the far-right hardliner minority control over which legislative proposals are even put up for a vote. This raises the possibility of a government shut-down due to running out of funds later this year; government shut-downs have become more frequent recently in the heavily politically polarized North American nation. There are also concerns that the country might default on its debt, bringing more political and economic instability to the region.

The troubled vote comes exactly two years after a violent coup attempt, supported by then-President, who refused to accept defeat in his bid for re-election. Armed militia storming the building where the legislative branch of the country’s government deliberates — United States Capitol — tried to stop the formal certification of the election’s result. The crisis was enabled partially by an outdated electoral system that often relies heavily on norms and custom in place of strict regulations.

Members of both chambers of the legislative branch, as well as then-Vice President of the country, had to be evacuated from the building. The attack resulted in several deaths.

Certain right-wing political figures, aligned with the former President, who had voiced their strong support for the armed insurgency perpetrating the coup and defended the organization (designated as “terrorist” in some countries) involved in organizing it have now been sworn-in as elected members of the lower legislative chamber, House of Representatives. They formed the core of the splinter group, leader of which had been accused of being involved in child sex trafficking and prostitution of minors.

The former President, who strives to maintain a strongman persona, is named in a number of investigations and criminal cases, ranging from tax evasion to stealing classified documents to inciting the coup attempt. He had also drawn accusations of nepotism after having appointed his daughter and son-in-law to important positions in his administration, and was embroiled in scandals involving, among others, paying off a porn star over an alleged affair.

The oil-rich North American country is struggling with high rate of gun violence (among the 20 most heavily affected countries in the world, according to a 2016 ranking), high cost of and difficult access to healthcare, lowest adult literacy rate in the region, and one of the highest incarceration rates in the world. These issues disproportionately affect communities of people of color, in no small part due to country’s economy having relied heavily on slavery in the past.

This post is inspired by Joshua Keating’s “If It Happened There” column in Slate, which I found to be as hilarious as it is illuminating. I can only wish someone would continue this kind of lighthearted yet much-needed work.

And so it has come to this. I finally quit Twitter… almost exactly a decade ago.

I could spin yarn and claim it was some major feat of clairvoyance, of course. That I foresaw all that happened lately with Twitter and decided to bail early. But it wasn’t that, really. I just felt strongly that centralized services are dangerous and unethical, and I decided to stop using them. Back then, Twitter was the last one on the chopping block for me.

The “Why”¶

Why did I feel so strongly about centralized services? Had you asked me ten years ago, it would have been difficult for me to explain. But even then I would have said it boils down to control and power dynamics. At the time I was a Free Software advocate, working for a Polish version of the FSFE. Software freedom was (and remains) important to me and it seemed obvious that one cannot have software freedom in a walled garden.

Hardly a strong, concrete argument, I know!

But it did turn out to be correct, didn’t it? It is about control and power dynamics. Anyone who migrated from Twitter to the Fediverse lately can attest to this. Anyone who read the Facebook Papers can understand this. Today, Twitter is run by an abuser, and Facebook is an abuser.

A bit of history¶

At the time I was using Diaspora and StatusNet (later renamed to GNU social), precursors to the Fediverse. Both were tiny, but I could see the value in the basic idea of decentralized social media: no single point of control, no single point of failure. I was also promoting that idea. Somewhat successfully, might I add, as I was able to convince the Polish Ministry of Administration and Digitization to create a StatusNet account. As far as I know this was the first official decentralized social media profile of a government institution — if that’s incorrect, I’d love to hear it!

One could claim that there is uninterrupted continuity between these older networks and the Fediverse. Friendica implements ActivityPub, the protocol that underpins the Fediverse today. It also implements Diaspora’s protocol, and the OStatus protocol that StatusNet used back in the day. Some Friendica instances had been running continuously for over a decade. They serve as a connection between the modern-day Fediverse and the decentralized social networks of years past.

This broader decentralized social network that morphed over the years into what we today call the Fediverse predated and outlived Google+. Diaspora was launched in November 2010. StatusNet is even older: it’s first and biggest public instance, identi.ca went live in July 2008. Google+ was launched in June 2014 and shuttered in April 2019.

Let’s stop here and ponder this extraordinary fact for a moment. Google, one of the largest and most influential tech companies in the world, threw all its weight behind Google+ — even going as far as outright forcing all YouTube users to use it. Yet, within five years from its inception Google+ was no more. And with it, all communities established and connections made on that platform.

Meanwhile, an open, decentralized social network, having none of the resources nor clout that Google+ had behind it, with no business model and no monetization scheme, happily carries on into its 15th year.

A Pet Peeve¶

Over the years I blogged a few times about decentralization of social media. Had some ideas on how to bring people over, suggested building decentralized protocols into the blogosphere (something that is now a reality with the WordPress ActivityPub plugin, for example).

A year after I had quit Twitter, in December 2013, I gave a talk at 30C3, where I called Twitter and Facebook “monopolies”. Back then that was a tough sell. Today, things seem different: the idea that social media platforms operated by Twitter and Meta are at least “monopoly-shaped” is acceptable and often accepted. That was not the last time I presented about the problem of centralization of the Internet.

Two years later, at the FSFE assembly during 32C3, I gave another talk about the insanity of having over fifty(!) different, incompatible open decentralized social networking protocols. Some of these slides did not age well, but the core point remains valid — compatibility is important, otherwise open decentralized social networks compete against each other and never get the chance to reach a size where the Network effect really kicks in for them.

Thankfully, now ActivityPub provides that common layer of compatibility for a lot of different projects. It’s important to keep that in mind; Mastodon might be the poster child of the Fediverse today, but software projects come and go. A healthy ecosystem of different but compatible software makes the whole network more resilient.

Privilege¶

I do recognize that my ability to quit Twitter cold turkey and still be able to find employment in my area of expertise, and to have a support network, is a form of privilege that not everyone has. But since I did have that privilege, even though I did not fully understand it at the time, I felt it was imperative I use it to do what I thought to be right: stop supporting centralized platforms like Twitter.

That’s the crux of it. By being on Twitter or Facebook, we support those platforms. And by not being anywhere else but on centralized platforms, we make it harder for other people to leave them — as we ourselves become one more person that those who do want to leave cannot find on the alternative networks, and thus one more reason to not stray outside the garden walls.

Not everyone has the privilege to quit Twitter. Not everyone has the spoons to set up an account on the Fediverse in addition to their Twitter presence. But many of us do, if we’re being honest with ourselves. The Twitter debacle shows why we should try to break out of the silos, if we do have that privilege and these spoons — in part for the sake of those who don’t.

Decentralized social networks in general, and the Fediverse in particular, are far from perfect. We should and will make them better.

By focusing our efforts on them, instead of on centrally controlled walled gardens, we can at least make sure that if we build something of value and import, if we create communities and connections there, if we invest the time and effort into setting up our presence, it will not all potentially disappear one day because a particular service got bought out or a megacorp got bored of running it.

This post was written for and originally published by the Institute of Network Cultures as part of the Dispatches from Ukraine: Tactical Media Reflections and Responses publication. It also benefited from copy editing by Chloë Arkenbout, and proofreading by Laurence Scherz.

Tackling disinformation and misinformation is a problem that is important, timely, hard… and, in no way new. Throughout history, different forms of propaganda, manipulation, and biased reporting have been present and deployed — consciously or not; maliciously or not — to steer political discourse and to goad public outrage. The issue has admittedly become more urgent lately and we do need to do something about it. I believe, however, that so far we’ve been focusing on the wrong parts of it.

Consider the term “fake news” itself. It feels like a new invention even though its literal use was first recorded in 1890. On its face it means “news that is untrue”, but of course, it has been twisted and abused to claim that certain factual reporting is false or manufactured — to a point where its very use might suggest that a person using it not being entirely forthright.

That’s the crux of it; in a way, “fake” is in the eye of the beholder.

Matter of trust¶

While it is possible to define misinformation and disinformation, any such definition necessarily relies on things that are not easy (or possible) to quickly verify: a news item’s relation to truth, and its authors’ or distributors’ intent.

This is especially valid within any domain that deals with complex knowledge that is highly nuanced, especially when stakes are high and emotions heat up. Public debate around COVID-19 is a chilling example. Regardless of how much “own research” anyone has done, for those without an advanced medical and scientific background it eventually boiled down to the question of “who do you trust”. Some trusted medical professionals, some didn’t (and still don’t).

As the world continues to assess the harrowing consequences of the pandemic, it is clear that the misinformation around and disinformation campaigns about it had a real cost, expressed in needless human suffering and lives lost.

It is tempting, therefore, to call for censorship or other sanctions against misinformation and disinformation peddlers. And indeed, in many places legislation is already in place that punishes them with fines or jail time. These places include Turkey and Russia, and it will surprise no one that media organizations are sounding alarms about them.

The Russian case is especially relevant here. On the one hand, the Russian state insists on calling their war of aggression against Ukraine a “special military operation” and blatantly lies about losses sustained by the Russian armed forces, and about war crimes committed by them. On the other hand, Kremlin appoints itself the arbiter of truth and demands that any news organizations in Russia propagate these lies on its behalf — using “anti-fake news” laws as leverage.

Disinformation peddlers are not just trying to push specific narratives. The broader aim is to discredit the very idea that there can at all exist any reliable, trustworthy information source. After all, if nothing is trustworthy, the disinformation peddlers themselves are as trustworthy as it gets. The target is trust itself.

And so we apparently find ourselves in an impossible position:

On one hand, the global pandemic, a war in Eastern Europe, and the climate crisis are all complex, emotionally charged high-stakes issues that can easily be exploited by peddlers of misinformation and disinformation, which thus become existential threats that urgently need to be dealt with.

On the other hand, in many ways, the cure might be worse than the disease. “Anti-fake news” laws can, just like libel laws, enable malicious actors to stifle truthful but inconvenient reporting, to the detriment of the public debate, and the debating public. Employing censorship to fight disinformation and misinformation is fraught with peril.

I believe that we are looking for solutions to the wrong aspects of the problem. Instead of trying to legislate misinformation and disinformation away, we should instead be looking closely at how is it possible that it spreads so fast (and who benefits from this). We should be finding ways to fix the media funding crisis; and we should be making sure that future generations receive the mental tools that would allow them to cut through biases, hoaxes, rhetorical tricks, and logical fallacies weaponized to wage information wars.

Compounding the problem¶

The reason why misinformation and disinformation spread so fast is that our most commonly used communication tools had been built in a way that promotes that kind of content over fact-checked, long-form, nuanced reporting.

According to Washington Post, “Facebook programmed the algorithm that decides what people see in their news feeds to use the reaction emoji as signals to push more emotional and provocative content — including content likely to make them angry.”

When this is combined with the fact that “[Facebook’s] data scientists confirmed in 2019 that posts that sparked [the] angry reaction emoji were disproportionately likely to include misinformation, toxicity and low-quality news”, you get a tool fine-tuned to spread misinformation and disinformation. What’s worse, the more people get angry at a particular post, the more it spreads. The more angry commenters point out how false it is, the more the algorithm promotes it to others.

One could call this the “outrage dividend”, and disinformation benefits especially handsomely from it. It is related to “yellow journalism”, the type of journalism where newspapers present little or no legitimate, well-researched news while instead using eye-catching headlines for increased sales, of course. The difference is that tabloids of the early 20th century didn’t get the additional boost from a global communication system effectively designed to promote this kind of content.

I am not saying Facebook intentionally designed its platform to become the best tool a malicious disinformation actor could dream of. This might have been (and probably was) an innocent mistake, an unintended consequence of the way the post-promoting algorithm was supposed to work.

But in large systems, even tiny mistakes compound to become huge problems, especially over time. And Facebook happens to be a gigantic system that has been with us for almost two decades. In the immortal words of fictional Senator Soaper: “To err is human, but to really foul things up you need a computer.”

Of course, the solution is not as simple as just telling Facebook and other social media platforms not to do this. What we need (among other things) is algorithmic transparency, so that we can reason about how and why exactly a particular piece of content gets promoted.

More importantly, we also need to decentralize our online areas of public debate. The current situation in which we consume (and publish) most of our news through two or three global companies, who effectively have full control over our feeds and over our ability to reach our audiences, is untenable. Monopolized, centralized social media is a monoculture where mind viruses can spread unchecked.

It’s worth noting that these monopolistic monocultures (in both the policy and software sense) are a very enticing target for anyone who would be inclined to maliciously exploit the algorithm’s weaknesses. The post-promoting algorithm is, after all, just software, and all software has bugs. If you find a way to game the system, you get to reach incredibly numerous audiences. It should then come as no surprise that most vaccine hoaxes on social media can be traced back to only 12 people.

Centralization obviously also relates to the ability of billionaires to just buy a social network wholesale or the inability (or unwillingness) of mainstream social media platforms to deal with abuse and extremism. They all stem from the fact that a handful of for-profit companies control the daily communication of several billion people. This is too few companies to wield that kind of power, especially when they demonstrably wield it so badly.

Alternatives do already exist. Fediverse, a decentralized social network, does not have a single company controlling it (and no shady algorithm deciding who gets to see which posts), and does not have to come up with a single set of rules for everyone on it (an impossible task, as former Twitter CEO, Jack Dorsey, admits). Its decentralized nature (there are thousands of servers run by different people and groups, with different rules) means that it’s easier to deal with abuse. And since it’s not controlled by a single for-profit company there is no incentive to keep bad actors in so as not to risk an outflow of users (and thus a drop in stock prices).

So we can start by at least setting up a presence in the Fediverse right now (following thousands of users who migrated there after Elon Musk’s Twitter bid). And, we can push for centralized social media walled gardens to be forced to open their protocols, so that their owners no longer can keep us hostage. Just like the ability to move a number between mobile providers makes it easier for us to switch while keeping in touch with our contacts, the ability to communicate across different social networks would make it easier to transition out of the walled gardens without losing our audience.

Media funding¶

As far as funding is concerned, entities spreading disinformation have at least three advantages over reliable media and fact-checking organizations.

First, they can be bank-rolled by actors who do not care if they turn a profit. Secondly, they don’t have to spend any money on actual reporting, research, fact-checking, and everything else that is both required and costly in an honest news outlet. Third, as opposed to a lot of nuanced long-form journalism, disinformation benefits greatly from the aforementioned “outrage dividend” — it is easier for disinformation to get the clicks, and create ad revenues.

Meanwhile, honest media organizations are squeezed from every possible side. Not the least by the very platforms that gate-keep their reach, or provide (and pay for) ads on their websites.

Many organizations, including small public grant-funded outlets, find themselves in a position where they feel they have to pay Facebook for “reach”; to promote their posts on its platform. They don’t benefit from the outrage dividend, after all.

In other words, money that would otherwise go into paying journalists working for a small, often embattled media organization, gets funneled to one of the biggest tech companies in the world, which consciously built their system as a “roach motel” — easy to get in, very hard to get out once you start using it — and now exploits that to extract payments for “reach”. An economist might call it “monopolistic rent-seeking”.

Meanwhile, the biggest ad network operator, Google, uses their similar near-monopoly position to extract an ever larger share of ad revenues, leaving less and less on the table for media organizations that rely on them for their ads.

All this means that as time goes by it gets progressively harder to publish quality fact-checked news. This is again tied to centralization giving a few Big Tech companies the ability to control global information flow and extract rents from that.

A move to non-targeted, contextual ads might be worth a shot — some studies show that targeted advertising offers quite limited gains compared to other forms of advertising. At the same time, cutting out the rent-seeking middle man leaves a larger slice of the pie on the table for publishers. More public funding (perhaps funded by a tax levied on the mega-platforms) is also an idea worth considering.

Media education¶

Finally, we need to make sure our audiences can understand what they’re reading, along with the fact that somebody might have vested interests in writing a post or an article in a particular way. We cannot have that without robust media literacy education in schools.

Logic and rhetoric have long been banished from most public schools as, apparently, they are not useful for finding a job. Logical fallacies are barely (if at all) covered. At the same time both misinformation and disinformation rely heavily on logical fallacies. I will not be at all original when I say that school curricula need to emphasize critical thinking, but it still needs to be said.

We also need to update the way we teach, to fit the current world. Education is still largely built around the idea that information is scarce and the main difficulty is acquiring it (hence its focus on memorizing facts and figures). Meanwhile, for at least a decade if not more, information is plentiful, and the difficulty lies in filtering it and figuring out which information sources to trust.

Solving the right problem, together¶

“Every complex problem has a solution which is simple, direct, plausible — and wrong”, observed H. L. Mencken. This describes the push for seemingly simple solutions to the misinformation and disinformation crisis, like legislation making disinformation (however defined) “illegal”, well.

News and fact-checking communities have limited resources. We cannot afford to spend them on ineffective solutions — and much less on in-fighting about proposals that are both highly controversial and recognized broadly as dangerous.

To really deal with this crisis we need to recognize centralization — of social media, of ad networks, of media ownership, of power over our daily communication, and in many other areas related to news publishing — and poor media literacy among the public as crucial underlying causes that need to be tackled.

Once we do, we have options. Those mentioned in this text are just rough ideas; there are bound to be many more. But we need to start by focusing on the right parts of the problem.

Disclaimer: I am not a lawyer. I am not your lawyer. None of this is legal advice. All of this might also be a horribly bad idea.

Ah, SEO link spam e-mails. If you have a blog that’s been online longer than, say, three years, you know what I’m talking about:

Hey,

I read your article at <link-to-a-blogpost-of-mine> talking about <actually-not-the-topic-of-the-blogpost>. I think your readers would benefit from a link to <link-to-an-irrelevant-or-trivial-piece>.

Would you consider linking to our article?

For a long time I just ignored these, flagging as spam and moving on. Obviously I am not going to link to some marketing crap that’s there only to drive up SEO of some random site.

But then that one spammer showed up in my mailbox, and he was persistent. Several e-mails and follow-ups within a month. I decided I needed a better strategy.

What if I told them to pay for a link being placed on my blog?

I asked for input on fedi, and after quite a few useful suggestions and comments, I drafted what is now my standard template to deal with these kinds of requests.

The Template¶

Hey,

thanks for reaching out. My going rate for a link placed on my blog is $500USD; I get to decide where and how I place it, and within what content. It will be placed in a regular blogpost, reachable by search engines, on the blog in question. It will stay up for at least a year. No other guarantees are made.

I require payment of half of the sum ($250, non-refundable) before I prepare the specific placement offer, for you to accept or reject. The placement, context and meaning of the link in the placement offer shall be determined at my sole and absolute discretion. There is no representation or warranty whatsoever as to whether the link is placed in a way that would imply an endorsement, or even fail to be an explicit or implied disparagement.

Once provided, the placement offer is final, and if rejected, I understand you are no longer interested in placing a link on my blog. At that point the initial payment is considered payment for my time and expertise in preparing the placement offer.

Once you accept the placement offer, I will put the link on-line within 10 business days, and I will expect payment in full at the latest 20 business days from it went online. After that period interest will accrue at 12% p.a., calculated annually.

Please be advised that any further communication from anyone at <company-name-or-domain-spam-e-mail-was-sent-from> or in relation to <domain-of-the-link-being-peddled> that is neither a clear rejection of this deal nor acceptance of the terms as outlined herein (and discussion about invoicing or accounting technicalities) will accrue a $50 processing fee. Any further communication from anyone at <company-name-or-domain-spam-e-mail-was-sent-from> or in relation to <domain-of-the-link-being-peddled>, including apparently unrelated to the matter at hand, amounts to acceptance of these terms, regardless of when it takes place and who the sender is. Any and all disputes must be subject only to the law of my jurisdiction (Iceland) and handled solely in the courts herein.

Do let me know if you have any specific invoicing/accounting requirements. I am looking forward to doing business with you.

The Point¶

The point, obviously, is to limit the amount of SEO link spam e-mails I have to deal with. But of course if somebody decides to take me up on the offer, I am happy to pocket the $500 to publish a blogpost about how they just paid $500 for the privilege of being made fun of, by me.

Yes, I will link to where they ask, yes it will be reachable by search engines, but also: yes, the link might have rel="sponsored nofollow" attribute set.

This is also somewhat the point of this very blogpost. Each and every SEO link spam e-mail claims that the sender “has read my site”. Well, if they did, they are now surely aware what’s in stock.

Finally, most SEO link spam e-mails mention you can “unsubscribe” by replying to them. I never “subscribed” to any of them in the first place, so that just feels wrong. More importantly though, I simply don’t trust the spammers to actually respect my request to be removed from their contacts database.

I do however trust that once they are informed that any further communication would cost them $50, they might not want to communicate further.

The Outcome¶

I have used the template several times over the last few months. I have not once heard back from any of the spammers that got served with it, and the overall amount of SEO link spam e-mails I receive seems to have gone down measurably — which might or might not be related to my use of the template, of course.

The Future¶

I would love to be able to charge SEO link spam e-mail senders even for the first e-mail they send me. So I am thinking of adding some kind of EULA to that effect to my blog.

I hate EULAs; I find the assumption that some terms are binding even if the visitor has not explicitly agreed to them (nor read them) to be asinine. But if that’s the world we live in, I might as well use it to make SEO link spam a bit more costly.

I would like to propose a new term: outrage dividend.

Outrage dividend is the boost in reach that content which elicits strong emotional responses often gets on social media and other content sharing platforms.

This boost can be related to human nature — an outrage-inducing article will get shared more. It can also be caused by the particular set-up of the platform a given piece of content is shared on — Facebook’s post-promoting algorithm was designed to be heavily biased to promote posts that get the “angry” reaction.

A tale of two media outlets¶

Imagine two media organizations.

A Herald is a reliable media organization, with great fact-checking, in-depth reporting, and so on. Their articles are nuanced, well-argued, and usually stay away from sensationalism and clickbaity titles.

B Daily is (for want of a better term) a disinformation peddler. They don’t care about facts, as long as their sensationalist, clickbaity articles get the clicks, and ad revenue rolls in.

Thanks to the outrage dividend, content produced by B Daily will get more clicks (and more ad revenue), as more people will engage with it simply because it’s exploiting our human nature; but it will also be put in front of more eyeballs because it causes people to be angry, and anger gets a boost (at least on Facebook).

Outrage Dividend’s compound interest¶

It gets worse: not only B Daily’s content is cheaper to produce (no actual reporting, no fact-checking, etc), not only does it get promoted more on the platform due to the particular angry reaction it causes in people, but also every time it gets fact-checked or debunked, that’s more engagement, and so even more reach.

Meanwhile, A Herald not only has to pay for expensive experts to do fact-checking, for reporters to do reporting, and so on, but also they feel they need to pay for reach, because their nuanced, in-depth, well-reasoned pieces get fewer clicks as they get promoted less by the platform’s algorithms.

Relation to tabloids / yellow journalism¶

There obviously is a relation here to yellow journalism and tabloids. I think it’s fair to say that these types of outlets use or exploit the outrage dividend for profit, basically basing their business model on it.

Of course, tabloid newspapers of (say) early 20th century did benefit from the human side of the outrage dividend (which made them possible and profitable in the first place). But the rise of global, centralized platforms like Facebook, with their content promoting algorithms that can apparently be gamed in order to reach effectively unlimited audiences, made the rift between how hard it is to get nuanced content reach a broad audience, and how easy it is to spread disinformation and misinformation, really problematic.

With all this in mind I think we need to seriously consider ways outrage dividend could be countered, and what options (technological, legislative, or other) are available for that.

I can’t believe I have to spell this out, but:
free/libre/open-source software developers and open web activists selflessly running independent services online are people too.

It seems this idea is especially difficult to grasp for researchers (including, apparently, whoever reviews and green-lights their studies). The latest kerfuffle with the Princeton-Radboud Study on Privacy Law Implementation shows this well.

“Not a human subject study”¶

The idea of that study seems simple enough: get a list of “popular” websites (according to the research-oriented Tranco list), send e-mails to e-mail addresses expected to be monitored for privacy-related requests (like privacy@example.com), and use that to assess the state of CCPA and GDPR implementation. Sounds good!

There were, however, quite a few problems with this:

• the e-mails that were sent out did not identify them as part of on-going research;
• the e-mails could easily be understood as a legal threat;
• the website admins in question were never asked if they wish to participate.

Imagine you’re running a small independent social media site and you get a lawyery-sounding e-mail about a privacy regulation you might not even have heard about, that ends with:

I look forward to your reply without undue delay and at most within 45 days of this email, as required by Section 1798.130 of the California Civil Code.

Should you reach out to a lawyer? That can easily get costly, fast. Is it okay to ignore it? That could end in an even costlier lawsuit. And so, now you’re losing sleep over something that sounds serious, but turns out to be a researcher’s idea of “not a human subject study”.

Humanity-erasure¶

The study’s FAQ consistently mentions “websites”, and “contacting websites”, and so on, as if there were no people involved in running them nor in answering these e-mails. Consider this gem (emphasis mine):

What happens if a website ignores an email that is part of this study?

We are not aware of any adverse consequences for a website declining to respond to an email that is part of this study. We will not send a follow-up email about an email that a website has not responded to, and we will not name websites when describing email responses in our academic research.

Sadly, nobody told this to the volunteer admin of a small social media site, who is perhaps still worrying (or even spending money on a lawyer) over this. But don’t worry, the Princeton University Institutional Review Board has determined that the “study does not constitute human subjects research”. So it’s all good!

This is not the first time such humanity-erasure happens, either. Some time ago, researchers at University of Minnesota conducted a study that involved submitting intentionally buggy patches to the Linux kernel.

They insisted that they were “studying the patching process”, but somehow missed the fact that that process involved real humans, many of whom volunteered time and effort to work on the Linux kernel. The developers were not amused.

Eventually, the researchers had to issue an apology for their lack of empathy and consideration for Linux kernel developers and their wasted time.

Tangent: taking “open” seriously¶

This is a bit tangential, but to me all this seems to be connected to a broader problem of people not treating communities focused on (broadly speaking) openness seriously.

In the case of the Princeton study, several Fediverse instance admins were affected. The University of Minnesota study affected Linux kernel developers. In both cases their effort (maintaining independent social media sites; developing an freely-licensed piece of software) was not recognized as serious or important – even if its product (like the Linux kernel) perhaps was.

I see this often in other contexts: people complain about Big Tech and “the platforms” a lot, but any mention of Fediverse as a viable alternative (both in the terms of a service, but also in terms of a funding model) is more often than not met with a patronizing dismissal. We’ve been seeing the same for years regarding free software, too.

Meanwhile, a proven abuser like Facebook can pull a Meta and everyone will dutifully debate how insightful and deep a move this is.

Oh, the humanity!¶

It is quite disconcerting that researchers seem unable to recognize the humanity of FLOSS developers or admins of small, independent websites or services. It is even more disturbing that, apparently, this tends to fly under the radar of review boards tasked with establishing if something is or isn’t a human-subject study.

And it is disgraceful to abuse scarce resources (such as time and energy) available to volunteer admins and FLOSS developers in order to run such inconsiderate research. It alienates a privacy-conscious, deeply invested community at a time when research into privacy and digital human rights is more important than ever.

A lot has already been written about different aspects of why most distributed blockchain-based consensus systems are just… bad. And yet we are still able to find new such reasons. At least I think this is a new one. I have not seen it mentioned anywhere so far.

Distributed blockchain-based consensus systems, as they are currently implemented, are an energy-waste ratchet.

I am specifically talking about systems like Bitcoin and Ethereum, and any other system that:

• is distributed;
• lets their users control some kind of “assets” by tying these to their “wallets” until they spend them;
• uses blockchain for consensus.

What’s in a wallet¶

When you have any assets on any such system, they are associated with some form of a wallet. That boils down to a file containing the private key, often password-protected, which needs to be stored somewhere safe. It is also necessary to have that file and the associated password in order to do anything with your assets.

We are, however, human, and as humans we are bad both at remembering passwords, and at keeping digital files safe for long periods of time. Passwords get forgotten. Harddrives fail or are thrown away.

And when that happens, there is no way to retrieve the assets in question. They’re lost, forever.

A wasteful ratchet¶

As time goes by and more people lose access to their wallets, more assets will be irretrievably lost. This is a one-way street, or in other words: a ratchet.

All those assets, now lost (like tears in… rain), nevertheless still took energy (sometimes an insane amount!) to mine or mint. Even if someone considers it worth it to use that energy on mining or minting in the first place, we can probably agree that for assets that get irretrievably lost, that energy has simply been wasted.

Mining capacity doesn’t go away with lost assets, though – and so, that (steadily growing most of the time) mining capacity is used to support transactions in a network with more and more assets that remain forever inaccessible.

Blockchain-based consensus systems inevitably waste energy on creating worthless, lost cryptoassets. With time, the amount of lost cryptoassets can only grow.

To make matters worse, for systems that are supply-limited (like Bitcoin) that also means that at some point the amount of lost cryptoassets will exceed the amount of still accessible ones.

About a week ago, @tomasino published a post on his contract-based dependency management idea (aka CBDM), and I would be lying if I said I didn’t like it.

Not only does it provide a better model for dependency management than SemVer or any other versioning scheme, but it also:

• provides strong incentive for developers to maintain extensive test suites for their own software;
• provides strong incentive for developers to help developers of their project’s dependencies maintain extensive test suites, too;
• provides very clear and unambiguous information on whether or not some functionality or behaviour of the dependency is, in fact, officially supported by dependency’s developers, or not;
• provides very clear and unambiguous information if some functionality or behaviour of the dependency has changed;
• makes it very, very clear who done goofed if a dependency upgrade breaks a dependent project.

What’s CBDM?¶

The basic idea boils down to this: when deciding if a given version of a given dependency is compatible with a dependent piece of software, instead of relying on version numbers – rely on tests that actually verify the functionality and behaviour that piece of software actually depends on.

In other words, when considering updating dependencies of a project, don’t look at version numbers, but look at tests of the dependency (and their results).

Tomasino’s post goes into more detail and is well-worth a read.

What’s wrong with version numbers?¶

Version numbers are are notoriously unreliable in predicting if something breaks after the upgrade. That’s the whole point of SemVer – to try to make them more reliable.

The problem is that it’s impossible to express, in a set of just few numbers, all the dimensions in which a piece of software might change. More importantly, certain changes might be considered irrelevant or minor by the developers, but might break projects that depend on some specific peculiarity.

Cue specifications, and endless debates whether or not a particular change breaks the specification or not.

How could CBDM work in practice?¶

Let’s say I’m developing a piece of software, call it AProject. It depends on a library, say: LibBee. LibBee developers are Gentlefolk Scholars, and therefore LibBee has quite extensive test coverage.

As the developer of AProject I specify the dependency not as:

LibBee ver x.y.z

…but as:

LibBee, (list of upstream tests I need to be unchanged, and to pass)

(Bear with me here and let’s, for the moment, wave away the question of how exactly this list of upstream tests is specified.)

This list does not need to contain all of LibBee’s tests – in fact, it should not contain all of them as that would effectively pin the current exact version of LibBee (assuming full coverage; we’ll get back to that). However, they should be tests that test all of LibBee’s functionality and behaviour AProject does rely on.

This set of tests becomes a contract. As long as this contract is fulfilled by any newer (or older) version of LibBee I know it should be safe for it to be upgraded without breaking AProject.

What if a LibBee upgrade breaks AProject anyway?¶

I say “should”, because people make mistakes. If upgrading LibBee breaks AProject even though the contract is fulfilled (that is, all specified tests have not been modified, and are passing), there is basically only a single option: AProject relied on some functionality or behaviour that was not in the contract.

That makes it very clear who is responsible for that unexpected breakage: I am. I failed to make sure the contract contained everything I needed. Thus a long and frustrating blame-game between myself and LibBee’s developers is avoided. I add the information about the additional test to the contract, and deal with the breakage as in any other case of dependency breaking change.

AProject just got a better, more thorough dependency contract, and I didn’t waste any time (mine nor LibBee developers’) blaming anyone for my own omission.

Win-win!

What if the needed upstream test does not exist?¶

If a test does not exist upstream for a particular functionality or behaviour of LibBee that I rely on, it makes all the sense in the world for me to write it, and submit it as a merge request to LibBee.

When that merge request gets accepted by LibBee’s developers, it clearly means that functionality or behaviour is supported (and now also tested) upstream. I can now add that test to AProject’s dependency contract. LibBee just got an additional test contributed and has more extensive test coverage, for free. My project has a more complete contract and I can be less anxious about dependency upgrades.

Win-win!

What if the needed test is rejected?¶

If LibBee developers reject my merge request, that is a very clear message that AProject relies on some functionality or behaviour that is not officially supported.

I can either decide to roll with it, still add that test to the contract, and keep the test itself in AProject to check each new version of LibBee when upgrading; or I can decide that this is too risky, and re-write AProject to not rely on that unsupported functionality or behaviour.

Either way, I know what I am getting into, and LibBee’s developers know I won’t be blaming them if they change that particular aspect of the library – after all, I’ve been warned, and have a test to prove it.

You guessed it: win-win!

Abolish version numbers, then?¶

No, not at all. They’re still useful, even if just to know that a dependency has been upgraded. In fact, they probably should be used alongside a test-based dependency contract, allowing for a smooth transition from version-based dependency management to CBDM.

Version numbers work fine on a human level, and with SemVer they carry some reasonably well-defined information. They are just not expressive enough to rely on them for dependency management. Anyone who has ever maintained a large project with a lot of dependencies will agree.

Where’s the catch?¶

There’s always one, right?

The difficult part, I think, is figuring out three things:

1. How does one “identify a test”?
2. What does it mean that “a test has not changed”?
3. How to “specify a test” in a dependency contract?

The answers to 1. and 2. will almost certainly depend on the programming language (and perhaps the testing framework used), and will almost certainly mostly define the answer to 3.

One rough idea would be:

1. A test is identified by it’s name (basically every unit testing framework provides a way to “name” tests, often requiring them to be named).
2. If the code of the test changes in any way, the test is deemed to have changed. Probably makes sense to consider some linting first, so that whitespace changes don’t invalidate the contracts of all dependent projects.
3. If a test is identified by it’s name, using that name is the sanest.

I really think the idea has a lot of merit. Software development is becoming more and more test-driven (which is great!), why not use that to solve dependency hell too?