Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Post-mortem?, published by gwern on August 5, 2022 on LessWrong.

As Covid is for the most part over and tests have not been in short supply for a very long time now, eliminating most of the rationale for complicated testing procedures meant to optimize test count/cost, it would be interesting for anyone who was involved in this to do a retrospective or post-mortem. Was pool testing useful? Why or why not? Can it be improved for future pandemics and should we care about it?

My impression, not having paid much attention to the details at the time (AI scaling was more important), is that various group/pool testing approaches were useful in a few niches, but were rendered useless in many places due to politics/organization/social flaws, and didn't apply to most places.

I read that use of pool testing was common in China early on in 2020 as part of the hotspot strategy to economize on testing extremely large groups of people where only a few would have active Covid. This is the most logical place to apply it, and it was a major government & societal priority, so that is a success.

I get the impression that they were able to switch to individual testing fairly early, in order to get faster turnaround and do more localized testing. But to the extent that pool testing was effective from the start, before testing manufacturing could scale up, and helped squash Covid when it was still squashable, then it was very valuable even if obsoleted.

Of course, this evaluation can change depending on how Covid ultimately shakes out in China. Was it part of a heavyhanded but ultimately highly effective Covid Zero policy which led to it turning into a hermit country enduring devastating economic fallout while merely postponed the inevitable variant which could defeat Covid Zero? In which case, pool testing was a tool which played a part in buying China entire years to vaccinate but its good was rendered useless when they bizarrely failed to make any good use of the time, dabbling in ineffective vaccines etc, and in practice, only doubling down repeatedly on coercive testing and surveillance until the martingale blew up in their face. Or was it part of a visionary resolute effort which spared China the mass death of the rest of the world as it came out the other side by persisting so long in country-wide lockdown until Covid just sorta stopped being a problem somehow?

Then its good was unalloyed and it's sad so few other countries could boast a better record (eg Taiwan; I dunno if they made any use of it).

Sewage/water testing in places like colleges: various universities or towns made heavy use of what is in effect pool testing to do mass screening of water to detect infections and then hotspot them. This was only useful as long as you have a baseline of ~0, otherwise testing water merely gives you a proxy for overall infections, which is a lot less useful than using it to detect & squash hotspots. In individual cases of orgs with great discipline, this apparently worked well, and avoided the need for constant individual testing (which was expensive, not possible early on, and still inadequate - as all the infections inside 'bubbles' show, just too much leakage, test error, incubation, idiosyncrasies and so on).

Since most of these places were in countries where Covid became endemic eventually anyway, and everyone got infected, and in cases like colleges the young students were the last to get vaccinated (because they are the least harmed) and so needed to buy a lot of time to avoid infection before vaccination (which probably didn't happen for most of them), these success stories seem to boil down to just buying time, which is a good deal less impressive.

At the margin, doesn't seem like improvements to pool testing would have helped much compared to other things like faster vaccination ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: $20K in Bounties for AI Safety Public Materials, published by ThomasWoodside on August 5, 2022 on The Effective Altruism Forum.

TLDR

We are announcing a $20k bounty for publicly-understandable explainers of AI safety concepts. We are also releasing the results of the AI Safety Arguments competition.

Background

Of the technologists, ML researchers, and policymakers thinking about AI, very few are seriously thinking about AI existential safety. This results in less high-quality research and could also pose difficulties for deployment of safety solutions in the future.

There is no single solution to this problem. However, an increase in the number of publicly accessible discussions of AI risk can help to shift the Overton window towards more serious consideration of AI safety.

Capability advancements have surprised many in the broader ML community: as they have made discussion of AGI more possible, they can also contribute to making discussion of existential safety more possible. Still, there are not many good introductory resources to the topic or various subtopics. If somebody has no background, they might need to read books or very long sequences of posts to get an idea about why people are worried about AI x-risk. There are a few strong, short, introductions to AI x-risk, but some of them are out of date and they aren’t suited for all audiences.

Shane Legg, a co-founder of DeepMind, recently said the following about AGI:

If you go back 10-12 years ago the whole notion of Artificial General Intelligence was lunatic fringe. People [in the field] would literally just roll their eyes and just walk away. [I had that happen] multiple times. [...] [But] every year [the number of people who roll their eyes] becomes less.

We hope that the number of people rolling their eyes at AI safety can be reduced, too. In the case of AGI, increased AI capabilities and public relations efforts by major AI labs have fed more discussion. Similarly, conscious efforts to increase public understanding and knowledge of safety could have a similar effect.

Bounty details

The Center for AI Safety is announcing a $20,000 bounty for the best publicly-understandable explainers of topics in AI safety. Winners of the bounty will win $2,000 each, for a total of up to ten possible bounty recipients. The bounty is subject to the Terms and Conditions below.

By publicly understandable, we mean understandable to somebody who has never read a book or technical paper on AI safety and who has never read LessWrong or the EA Forum. Work may or may not assume technical knowledge of deep learning and related math, but should make minimal assumptions beyond that.

By explainer, we mean that it digests existing research and ideas into a coherent and comprehensible piece of writing. This means that the work should draw from multiple sources. This is not a bounty for original research, and is intended for work that covers more ground at a higher level than the distillation contest.

Below are some examples of public materials that we value. This should not be taken as an exhaustive list of all existing valuable public contributions.

AI risk executive summary (2014)

Concrete Problems in AI Safety (2016)

Robert Miles’ YouTube channel (2017-present)

AGI Safety From First Principles (2020)

The case for taking AI risk seriously as a threat to humanity (2020)

Unsolved Problems in ML Safety (2021)

X-risk Analysis for AI Research (2022)

Note that many of the works above are quite different and do not always agree with each other. Listing them isn’t to say that we agree with everything in them, and we don’t expect to necessarily agree with all claims in the pieces we award bounties to. However, we will not award bounties to work we believe is false or misleading.

Here are some categories of work we believe could be valuable:

Exe...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Rant on Problem Factorization for Alignment, published by johnswentworth on August 5, 2022 on The AI Alignment Forum.

This post is the second in what is likely to become a series of uncharitable rants about alignment proposals (previously: Godzilla Strategies). In general, these posts are intended to convey my underlying intuitions. They are not intended to convey my all-things-considered, reflectively-endorsed opinions. In particular, my all-things-considered reflectively-endorsed opinions are usually more kind. But I think it is valuable to make the underlying, not-particularly-kind intuitions publicly-visible, so people can debate underlying generators directly. I apologize in advance to all the people I insult in the process.

With that in mind, let's talk about problem factorization (a.k.a. task decomposition).

HCH

It all started with HCH, a.k.a. The Infinite Bureaucracy.

The idea of The Infinite Bureaucracy is that a human (or, in practice, human-mimicking AI) is given a problem. They only have a small amount of time to think about it and research it, but they can delegate subproblems to their underlings. The underlings likewise each have only a small amount of time, but can further delegate to their underlings, and so on down the infinite tree. So long as the humans near the top of the tree can “factorize the problem” into small, manageable pieces, the underlings should be able to get it done. (In practice, this would be implemented by training a question-answerer AI which can pass subquestions to copies of itself.)

At this point the ghost of George Orwell chimes in, not to say anything in particular, but just to scream. The ghost has a point: how on earth does an infinite bureaucracy seem like anything besides a terrible idea?

“Well,” says a proponent of the Infinite Bureaucracy, “unlike in a real bureaucracy, all the humans in the infinite bureaucracy are actually just trying to help you, rather than e.g. engaging in departmental politics.” So, ok, apparently this person has not met a lot of real-life bureaucrats. The large majority are decent people who are honestly trying to help. It is true that departmental politics are a big issue in bureaucracies, but those selection pressures apply regardless of the peoples’ intentions. And also, man, it sure does seem like Coordination is a Scarce Resource and Interfaces are a Scarce Resource and scarcity of those sorts of things sure would make bureaucracies incompetent in basically the ways bureacracies are incompetent in practice.

Debate and Other Successors

So, ok, maybe The Infinite Bureaucracy is not the right human institution to mimic. What institution can use humans to produce accurate and sensible answers to questions, robustly and reliably? Oh, I know! How about the Extremely Long Jury Trial? Y’know, because juries are, in practice, known for their extremely high reliability in producing accurate and sensible judgements!

“Well,” says the imaginary proponent, “unlike in a real Jury Trial, in the Extremely Long Jury Trial, the lawyers are both superintelligent and the arguments are so long that no human could ever possibility check them all the way through; the lawyers instead read each other’s arguments and then try to point the Jury at the particular places where the holes are in the opponent’s argument without going through the whole thing end-to-end.”

I rest my case.

Anyway, HCH and debate have since been followed by various other successors, which improve on their predecessors mostly by adding more boxes and arrows and loops and sometimes even multiple colors of arrows to the diagram describing the setup. Presumably the strategy is to make it complicated enough that it no longer obviously corresponds to some strategy which already fails in practice, and then we can bury our heads in the sand and pretend ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Counterfactuals are Confusing because of an Ontological Shift, published by Chris Leong on August 5, 2022 on The AI Alignment Forum.

I recently became unstuck on counterfactuals. I now believe that counterfactuals are confusing, in large part, because they entail preserving our values through an ontological shift.

In our naive ontology, when we are faced with a decision, we conceive of ourselves as having free will in the sense of there being multiple choices that we could actually take. These choices are conceived of as actual and we when think about the notion of the "best possible choice" we see ourselves as comparing actual possible ways that the world could be. However, we when start investigating the nature of the universe, we realise that it is essentially deterministic and hence that our naive ontology doesn't make sense. This forces us to ask what it means to make the "best possible choice" in a deterministic ontology where we can't literally take a choice other than the one that we make.

This means that we have to try to find something in our new ontology that roughly maps to our old one. For example, CDT pretends that at the point of the decision that we magically make a decision other than that which we actually end up making. For example, we were intending to turn left, but at the point of the decision we're magically altered to turn right instead. This works well for many scenarios, but notably fails for Newcomb's problem and similar. Here updateless counterfactuals are arguably a better fit since they allow us to model the existence of a perfect predictor. However, there is still too much uncertainty about how they should be constructed for this answer to be satisfactory.

Given that we're trying to map a notion onto another ontology that doesn't natively support it, it isn't that surprising there isn't an obvious, nice, neat way of doing it.

Newcomb's problem exposes the tension between the following two factors:

a) wanting to ensure that the past is the same in each counterfactual to ensure that they are comparable with each other, in the sense of it being fair to compare the counterfactuals in order to evaluate the decisionb) maintaining certain consistency conditions that are present in the problem statement.

The tension is as follows: If we hold the past constant between counterfactuals then we've violated the consistency conditions as we have a point in time when the agent takes a decision despite the fact that this is inconsistent with the previous time, plus the laws of physics. However, if we backpropogate the impacts of the agent's choice through time to create a consistent counterfactual, it's unclear whether the two counterfactuals are comparable since the agent is facing different scenarios, in terms of how much money is in the box.

It turns out that Newcomb's Problems is more complicated than I previously realised. In the past, I thought I had conclusively demonstrated that we should one-box by refuting the claim that one-boxing only made sense if you accepted backwards causation. However, this now seems insufficient as I haven't explained why we should maintain the consistency conditions over comparability after making the ontological shift.

In the past, I might have said that these consistency conditions are what define the problem and that if we dropped them it would no longer be Newcomb's Problem. However, that seems to take us towards a model of counterfactuals being determined by social convention, which only seems useful as a descriptive model, not a prescriptive model. My current approach now tends to put more focus on the evolutionary process that created the intuitions and instincts underlying these incompatible demands as I believe that this will help us figure out the best way to stitch them together.

In any case, understanding...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: The Falling Drill, published by Screwtape on August 5, 2022 on LessWrong.

Summary: The Falling Drill is intended to practice being wrong in a comfortable practice environment, intended to help us handle being wrong out in the world easier and with less stress.

Tags: Small, repeatable

Purpose: The first thing you learn in martial arts is the ability to fall. The corresponding rationalist skill is the ability to realize that you are wrong. Since modern western society (and possibly most human societies) discourage admitting you’re wrong, it can help to do it repeatedly and get used to it.

Materials: A device that can connect to wikipedia. A list of partial statements such as “The population of Boston is. . .” and “the melting point of mercury is. . .” A suggested list is here, and we suggest writing the questions down on individual cards before the meetup so each person only sees one card at a time.

Announcement Text: One of the most important parts of intellectual progress is learning to change your mind. The first step of changing your mind is realizing that you were wrong about something. Today we're going to practice that often painful realization, in a small way and in a low pressure situation. Without this skill, how can you debate an important issue or confront a challenging topic? You might argue long after it's clear to others that you've lost, because admitting it feels like defeat.

The first lesson any martial artist learns is often how to fall. There are ways to make the landing easier, but they all start from the knowledge that falling isn't the worst thing in the world.

Description:

1. Read one of the statements, then complete it as best you can. "The population of Boston is four hundred thousand."

2. Look it up on wikipedia. If you’re wrong, then announce to the room “I was wrong about the population of Boston. It's over six hundred thousand.”

3. Hand the questions to the next person in the circle, and the process begins again.

Notes: Some minor variations to play with: Practice saying “I don’t know” when you’re asked the question. Practice saying your answer loudly and confidently, such as "The population of Boston is four hundred thousand! Only an idiot would think it was higher than five!" It is good to know what you do not know, but it is also important to be able to back down from a strong claim. In particular, I suggest spending the most time practicing the version that you’re the least comfortable with. Harder variations involve the other participants giving fake mockery for being wrong, changing your mind, or "flip flopping." I strongly discourage doing that the first time you run The Falling Drill with a particular group.

Feedback:

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: [link post] The Case for Longtermism in The New York Times, published by abierohrig on August 5, 2022 on The Effective Altruism Forum.

An adapted excerpt from What We Owe The Future by Will MacAskill is now live in The New York Times and will run as the cover story for their Sunday Opinion this weekend.

I think the piece makes for a great concise introduction to longtermism, so please consider sharing the piece on social media to boost its reach!

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Announcing the SPT Model Web App for AI Governance, published by Paolo Bova on August 4, 2022 on The Effective Altruism Forum.

Modeling Cooperation is excited to announce our web app SPT Model—an interactive tool to explore a model of AI competition built in collaboration with Robert Trager.

Will AI safety breakthroughs always lead to safer AI systems?

Before long, we may be capable of creating AI systems with high levels of general intelligence. Such capable systems may present considerable risks when misaligned. Yet, suppose that thanks to incredible work by the AI safety community, we make a breakthrough which enables us to align many (but not all) AI systems and make them act according to our values at a relatively low performance cost. We also communicate this research to all relevant companies and governments. Everyone knows how to make their AI systems safer.

Fast forward several years and we may find that despite these efforts, the AI systems that companies deploy are just as risky as they were before the breakthrough. Our question to the reader is: why did this happen?

Announcing the web app

One way of looking at this question is to explore the safety-performance tradeoff model (SPT Model) of AI competition created by Robert Trager, Paolo Bova, Nicholas Emery-Xu, Eoghan Stafford, and Allan Dafoe.

At Modeling Cooperation, we worked with Robert Trager to implement this model in an interactive web app we call SPT Model.

The web app lets you explore a model of a competition to build emerging technologies. The model shows how failure scenarios like the one above can emerge, and there are many more insights contained within.

We built this tool as we believe that when trying to understand a model, there is no substitute for exploring it yourself. Our web app lets you set the parameters for two different scenarios and compare the effects graphically.

Answering the question

Let’s return now to the puzzle we presented above. What obstacles prevent AI safety breakthroughs from improving the safety of AI systems?

We will provide an answer below. Before that, you may wish to first experiment with the web app.

One phenomenon at play in the puzzle above is “risk compensation”.

Risk compensation occurs when making an activity safer motivates people to take more risk. Most times, risk compensation erodes but does not eliminate the benefits of safety technologies (think of seatbelts and bicycle helmets, for example). However, for emerging technologies, such as AI, a key difference is competition. Companies will take on more risk to outcompete their rivals.

We visualize these effects of a safety breakthrough in the safety-performance tradeoff plot below. The safety-performance tradeoff plots the maximum levels of performance we can achieve for each level of safety. If companies accept a higher level of risk, then conditional on avoiding a disaster, they can expect to produce a more capable AI system than otherwise and so reap a higher benefit. Such a scenario is analogous to the risk-return tradeoff employed in stock-market investing. It seems reasonable to expect companies and other institutional actors to consider a similar tradeoff when investing in new, potentially risky, strategic technologies.

In the puzzle we presented above, the breakthrough means if firms invest in AI systems with the same performance, they will be safer than before (see the upward arrow in Figure 2). However, this breakthrough has also enabled companies to take on even riskier AI systems to outcompete their rivals (the declining arrow in Figure 2). This competition can completely erode the benefits of the safety insight.

Of course, this isn’t always true. We now turn to a scenario where AI companies operate at such a high level of performance that attaining higher performance is prohibitively costly ...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Bridging Expected Utility Maximization and Optimization, published by Daniel Herrmann on August 5, 2022 on The AI Alignment Forum.

Background

This is the second of our (Ramana, Abram, Josiah, Daniel) posts on our PIBBSS research. Our previous post outlined five potential projects that we were considering pursuing this summer. Our task since then has been to make initial attempts at each project. These initial attempts help us to clarify each project, identify the primary problems that need to be solved, and perhaps discover a promising line of attack or two.

This post is aimed at the second proposal from our previous post. There we asked: what is the connection between an agent that maximizes expected utility and an agent that succeeds in action? Here we will outline a few of the problems we see in this area and potential routes for solving them.

Expected Utility Maximization and Forming Expectations

In economics and formal philosophy, the standard characterization of a rational agent is an agent who maximizes expected utility. Informally, when such an agent has a set of options available to her, she chooses the one that maximizes the expectation of her utility function, where the expectation is taken relative to her subjective degrees of belief. The claim that expected utility (EU) maximization characterizes rationality is usually supported by representation theorems (see here for a good, quick introduction). Expected utility maximization plays a core role in the philosophical foundations of decision theory, game theory, and probabilism. Given that EU maximization plays such a central role in theories of rationality, and given that there is a vast literature surrounding it, it seems very plausible that EU maximization would help us think precisely about agency.

Despite this, it seems that that expected utility theory doesn’t seem to predict anything (or, at the very least, you need to combine EU maximization with certain complexity notions to get something that is weakly predictive). Obviously this is an issue, given that we want notions of agency to constrain our expectations about the behaviour and effects of concrete systems. This is the sense in which we want to bridge the gap between expected an agent’s utility maximization and its success in action (or, rather, our expectation that it will be successful in action). If we know that a (sophisticated) EU maximizer has a certain goal, then we should be able to infer something about the likely unfolding of the world.

The primary obstacle for making EU maximization predictive is that for many systems we can reverse engineer a utility function and a probability distribution such that the system’s behavior is maximizing expected utility relative to that utility / probability pair. For example, consider a rock. We can say that the rock’s highest preference is to just sit there (or to roll, if it gets bumped). Voilà, a maximizer.

One way of understanding what went wrong here is that the system we started with (the rock) has no obvious utility scale (what John Wentworth calls a “measuring stick of utility”, but which a measurement theorist would call a “representation”). An obvious utility scale is something like money, or food. If someone sees me pay $1 to exchange an apple for an orange and $1 to exchange an orange for an apple, then EU theory says either (1) I’m irrational, or (2) I don’t have a utility function over apples and oranges, or (3) I don’t value having more money given the same fruit. These aren’t particularly strong requirements, but they are requirements nonetheless.

In order to understand conditions under which EU maximization can help us form useful expectations about the world, we wanted to identify some models that do seem to help us form expectations about the world going in a way that seems sensiti...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Jan Leike: On the windfall clause, published by Cullen OKeefe on August 4, 2022 on The Effective Altruism Forum.

Jan wrote this thoughtful critique of the Windfall Clause back in 2020, and I thought it should be posted publicly. Posted with Jan's permission. Key excerpt:

A core challenge when trying to design a windfall clause is that there is an incredibly strong incentive to find a loophole once the clause takes effect. If you run an organization who signed a windfall clause and in the future the unlikely comes to pass and you actually end up making $11 trillion in annual profits, it would be rational for you to spend up to $10 trillion on legal fees to try to get out of that clause just for that year; preferably in a way that doesn't cost you too much credibility. Companies are doing this already–this is why the big internet companies pay hardly any taxes.

So how can we implement a windfall clause so that it withstands a multi-trillion dollar search for loopholes? This is incredibly hard, because we need to write and implement the clause today, before the windfall happens (which might never happen at all). This means it has to be done on a budget that is many orders of magnitude smaller than the actual windfall. Let's call this asymmetry the legal offense/defense ratio: We have to defend the "spirit" of the windfall clause against a future version of the organization (possibly with different people in charge) who is experiencing a windfall and wants to get out of the clause. So if you tried to figure out the implementation of a legally tight windfall clause on an insanely high budget like 10 million dollars to safeguard a $10 trillion windfall, you need to withstand a legal offense/defense ratio of at least 1:1,000,000.[1]

These numbers are certainly made up, but let's roll with them and look at some historical precedent: What are legal battles where one party has spent that much more money than the other? I'm certainly not a lawyer, but I want to highlight two examples.

The Tobacco Master Settlement Agreement was one of the highest stakes legal cases in history: a bunch of US states sued the big tobacco companies for the increased burden of long-time smokers on the healthcare system. The tobacco companies ultimately lost and were sentenced to pay $206 billion over 25 years.[2] Because of incentive fees, the states' legal fees exceeded $8 billion. I don't know how much the tobacco companies spent on legal fees, but in order to get above the 1:1,000,000 ratio, they would have needed to keep their expenses below $8,000. We can safely assume that they spent much more than that; $8,000 probably wouldn't even enable you to attend most of the court sessions even if you are defending yourself, live right next to the court, and pay yourself minimum wage.

This may sound ridiculous, but this was almost the case in the famous "McLibel" case: Two people were sued by MacDonald's in 1990 for libel because they distributed some flyers accusing the corporation of various malpractices: destroying rainforests, mistreating animals–the usual stuff. One of the defendants was a part-time bar-worker earning a maximum of £65 a week, and the other one was an unemployed postal worker. The two defendants were denied free legal aid because apparently the UK libel system is not great at this. Therefore they represented themselves, plus spending about £30,000 over ten (!) years, not counting their personal time and various pro bono legal advice. In contrast, MacDonald's allegedly spent several million dollars on the case. Still, this is a legal offense/defense ratio of below 1:1,000. The two defendants ultimately lost the case and had to pay £40,000 to MacDonald's, but the Judge also upheld some of their claims.

The European Court for Human Rights later ruled that they weren't given a fair tri...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Fiber arts, mysterious dodecahedrons, and waiting on “Eureka!”, published by eukaryote on August 4, 2022 on LessWrong.

Part 1: The anomaly

This story starts, as many stories do, with my girlfriend 3D-printing me a supernatural artifact. Specifically, one of my favorite SCPs, SCP-184.

We had some problems with the print. Did the problems have anything to do with printing a model of a mysterious artifact that makes spaces bigger on the inside, via a small precisely-calibrated box? I would say no, there’s no way that be related.

Anyway, the image used for the SCP in question, and thus also the final printed model, is based a Roman dodecahedron. Roman dodecahedrons are a particular shape of metal object that have been dug up in digs from all over the Roman period, and we have no idea why they exist.

Many theories have been advanced. You might have seen these in an image that was going around the internet, which ended by suggesting that the object would work perfectly for knitting the fingers of gloves.

There isn’t an alternative clear explanation for what these are. A tool for measuring coins? A ruler for calculating distances? A sort of Roman fidget spinner? This author thinks it displays a date and has a neat explanation as for why. (Experimental archaeology is so cool, y’all.)

Whatever the purpose of the Roman dodecahedron was, I’m pretty sure it’s not (as the meme implies is obvious) for knitting glove fingers.1

Why?

1: The holes are always all different sizes, and you don’t need that to make glove fingers.

2: You could just do this with a donut with pegs in it, you don’t need a precisely welded dodecahedron. It does work for knitting glove fingers, you just don’t need something this complicated.

3: The Romans hadn’t invented knitting.

Part 2: The Ancient Romans couldn’t knit

Wait, what? Yeah, the Romans couldn’t knit. The Ancient Greeks couldn’t knit, the Ancient Egyptians couldn’t knit. Knitting took a while to take off outside of the Middle East and the West, but still, almost all of the Imperial Chinese dynasties wouldn’t have known how. Knitting is a pretty recent invention, time-wise. The earliest knit objects we have are from Egypt around 1000 CE.

This is especially surprising because knitting is useful for two big reasons:

First, it’s very easy to do. It takes yarn and two sticks and children can learn how. This is pretty rare for fabric manufacturing – compare, for instance, weaving, which takes an entire loom.

Sidenote: Do you know your fabrics? This next section will make way more sense if you do.

Woven fabricKnit fabricCommonly found in: trousers, collared/button up shirts, bedsheets, dish towels, woven boxers, quilts, coats, etc. Not stretchy. Loose threads won’t make the whole cloth unravel.Commonly found in: T-shirts, polo shirts, leggings, underwear, anything made of jersey fabric, sweaters, sweatpants, socks.Stretchy. If you pull on a lose thread, the cloth unravels.

Second, and oft-underappreciated, knitted fabric is stretchy. We’re spoiled by the riches of elastic fabric today, but it wasn’t always so. Modern elastic fabric uses synthetic materials like spandex or neoprene; the older version was natural latex rubber, and it seems to have taken until the 1800s to use rubber to make clothing stretchy. Knit fabric stretches without any of those.

Before knitting, your options were limited – you could only make clothing that didn’t stretch, which I think explains a lot of why medieval and earlier clothing “looks that way”. A lot of belts and drapey fabric. If something is form-fitting, it’s probably laced. (.Or just more-closely tailored, which unrelatedly became more of a thing later in the medieval period.)

You could also use woven fabric on the bias, which stretches a little.

Medieval Europe made stockings from fabric cut like this. Imagine a sock ma...