Taking a day off work is not a strike.

Power will only recognize threats to power, and strikes threaten a company’s bottom line. If everyone takes a day off work, the effect on production can be completely eliminated by the elasticity of work. This is especially true for exempt¹ employees, who are usually expected to deliver regardless of time off and have been indoctrinated over their careers into making up for “lost time”. In fact, the evolution of the economy during the era of modern trade unions has made strike effectiveness more complicated—a company can weather a strike for longer if it has spread its production out across demographics² that aren’t likely to participate. The effectiveness of an employee withholding their labour is basically measured by their contribution to the marginal cost of the product and that product’s place in the economy. This is evidenced at the extremes by how power attempts to limit strikes in key industries, from no-strike clauses in healthcare, the Railway Labor Act,³ and the union-busting of air traffic controllers.⁴

But that doesn’t mean that a daylong walkout—like what we’re seeing in Minneapolis today—won’t be effective. It’s just not a strike; it’s a protest.

While it’s easy to be tricked into thinking that nonviolent protests don’t work, the reality is that they are a visible shot across the bow of power. Protests are a reversion to direct democracy, a show of numbers by people united by a common grievance when their elected representatives fall short. They bring like-minded people together.

Most importantly though, protests are a threat of escalation. The “3.5% rule” in political science is intended to assess the credibility of that threat, but it’s not founded on any specific principles that make it inevitable. If you’re unhappy with things, get out there and meet people who share your view. Plan actions together that will threaten power in ways that encourage better outcomes.

Protests are collective action. Which makes them union as fuck.

I bought mine in 2022 when my employer instituted a mandatory return to office policy.¹ Moving to Oakland meant that my new commute was substantially more dangerous, and having a kid meant it was more important than ever that I not die. Nor, ideally, be maimed.

There are a lot of options on the market, which raises a lot of questions. The answers depend on what you’re optimizing for. Airbag systems are available with mechanical or electronic triggers. They may be integrated into gear, worn underneath, or worn on top. Different models cover different parts of the body. Here’s how I chose:

External, Internal, or Integrated?

Broadly, there are three different kinds of air bag garments:

1. “External” vests you wear over your jacket/suit
2. “Internal” vests you wear under your jacket/suit
3. “Integrated” systems built into your jacket/suit

Integrated solutions have a lot going for them and are worth considering if you’re shopping for a complete set of gear, but I’m very happy with my Roadcrafter so for me the question was just “external” vs “internal”.

Before the pandemic I rode my motorcycle every single day, which transfers a lot of sweat and grime into riding gear. My helmet’s pads and my whole suit (sans armour) are machine-washable, but at the time none of the internal airbags on the market had a better story than “use a moist rag and don’t get the inflator wet”. Nowadays some internal vests can be broken down for cleaning, but they still come with other caveats like “your jacket must have at least 4cm of extra space around the chest” which foreshadow problems² that I don’t don’t need in my life.

Meanwhile, the bladder of an external vest is not constrained, free to fill the gap between my shoulders and helmet. Nor, practically speaking, does it need to be cleaned. It doesn’t even add an extra step to gearing up; at this point my suit wears the vest and I don (or doff) them together as one.

Mechanical or Electronic Trigger?

As far as I’m aware only external vests and some integrated systems offer mechanical triggers, which connect to a tether attached to the motorcycle. These are very simple systems—yanking on the tether with around 60 lbs of force triggers the inflator, which does its job in the blink of an eye.

Electronically-triggered airbags use advanced machine learning algorithms fed by GPS data and multiple IMUs. Some can communicate with an additional IMU on the motorcycle,³ some support multiple deployments before the inflator is discharged, and they can all react (and deploy) before the rider even realizes they’re crashing.

They also don’t work when the battery is dead and require factory service after discharge. Many models only fire at GPS speeds greater than 15 mph⁴ and some manufacturers engage in pretty unsavoury pricing practices.

I chose the tether. I like that I can check if it’s “on” by giving the tether a tug. I’d gone down with it once before, but the airbag didn’t deploy because I stayed on the bike, having lowsided at low speed in a mud patch. This was a fine outcome for me because there was also no real threat of injury—I even had enough time to kill the engine on the way down—but if it had deployed, the propellant is a standard threaded 60 cc CO₂ cartridge and the spring-loaded trigger mechanism can be reset on the side of the road with an Allen key.

It’s easy to imagine downsides to mechanical tethers, but they don’t really come up in practice. I’ve figured out how to connect the buckle with one hand while I’m riding,⁵ but rarely forget to do it before taking off. If I forget to unbuckle when dismounting, the trigger’s minimum force provides feedback that’s pretty difficult to ignore.⁶ One company in Europe uses an inertial reel for their tether (instead of a nylon-webbed bungee), which guarantees the fastest possible deployment for a mechanical system without compromising the rider’s ability to stand up on the pegs.

Coverage

When I was first shopping, internal vests were sized for fitting under jackets, which limited their coverage. External vests extended further, with the bladder overlapping the rider’s hips and tailbone. Nowadays maximalists that don’t mind spending more time getting dressed can choose internal (or integrated) airbag systems with coverage all the way out to the rider’s wrists and thighs.

Collarbone fractures are the most common injury among riders wearing passive protection. Hip fractures are less common, but affect your mobility for the rest of your life. My goal was not to prevent all injuries, but to minimize long-term impacts. Your goals may be different—and it’s totally reasonable to want maximum coverage—but past a certain point optimizing for this favours internal or integrated systems; if you’re inclined towards a mechanical trigger, you may have to compromise.

Anything Else?

If you made it this far, I chose the Helite Turtle 2. The fabric around its back armour has seen better days, but it’s still serviceable and I’m still wearing it on the commute. I’m planning on buying a Helite H-MOOV as soon as someone will sell one to me.

In 2021 Ryan F9 posted a pretty good video about airbags (he also preferred the mechanical Helite) and in 2024 he posted another (same). Note that he’s not an ATGATT guy, and that his (and my) perspective is focused on the hazards of street riding. Maybe you’re shopping for the track. Maybe you think bulky vests with tethers look dorky.⁷ Everyone’s needs are different and ultimately my choices may be different from yours, but hopefully reading this can help you decide what’s best for you. All I can say for sure is that crashing with an airbag is wayy better than crashing without one.

This post is not readable in the feed due to a bug, follow this link to check it out in your browser!

Sell enough that you’re not crying in the shower if the company goes out of business tomorrow, but not so much that you’re crying in the shower if the stock goes up 100⨉ in ten years.

So I sold enough to become debt-free, and got fired shortly after that². I bought up what I could afford in the following weeks³, SQ did make it to 100⨉ above my strike in 2021, and at no point was I crying in the shower over my investment decisions!

Best advice I ever got.

This post is not readable in the feed due to a bug, follow this link to check it out in your browser!

My commute home from work did not go as planned.

This post is not readable in the feed due to a bug, follow this link to check it out in your browser!

I’m of the opinion that you should never use mmap, because if you get an I/O error of some kind, the OS raises a signal, which SQLite is unable to catch, and so the process dies. When you are not using mmap, SQLite gets back an error code from an I/O error and is able to take remedial action, or at least compose an error message.

I have ranted about mmap¹ on this site before, and avoidable crashes ought to be a compelling reason to avoid it, but people still use it! Perhaps it would help to ask why:

Performance

This is the usual reason people give. “mmap is faster” is usually supported by two reasons: syscall overhead and buffer copying.

Syscalls were an uncommon bottleneck twenty years ago, and overhead for common syscalls (like pread) have decreased significantly since then. These days their overhead is a rounding error compared to the work they perform.

Likewise, the people in userland whose performance depends on zero-copy techniques are writing filesystems, not using them². VM CoW is good enough for the remaining people who can express their work units as some multiple of the VM’s page size.

Shared Memory

SQLite uses mmap³ for manifesting path-addressible garbage-collected interprocess shared memory regions, which is the only reason I’ve found for which there is no other option. A POSIX shmem can’t be reliably associated with a database when the process is chrooted and will leak if the last process using it dies without releasing its claim first, which is super common on platforms with aggressive oom behaviour (like iOS).

So this is where I step back from the doomerism a bit. Practically speaking, the combination of file monitoring via DISPATCH_SOURCE_TYPE_VNODE and the strategic use of mach_vm_read_overwrite when significant time has elapsed since the last access has, in practice, almost entirely eliminated mmap-related crashes in SQLite (the API will return SQLITE_IOERR_VNODE (6922) instead) with no measureable performance cost. But these mitigations are so much more complicated than “an API that returns an error if it fails”, which brings us finally to the real reason people use mmap,

Convenience

mmap is from a time before APIs expressed their own opinions intentionally⁴ and its ease of use makes it an extremely attractive nuisance. Nobody likes to admit to being lazy, but for files of arbitrary size (like databases), jumping around with pointer arithmetic is hard to beat: one bounds check⁵ and no memory management. Humans follow incentives, and C provides an environment where it’s way easier to ignore errors than to handle them. Consider what SQLite does instead, roughly 8,000 lines of code implementing a pagerhttps://www.sqlite.org/src/file?udc=1&ln=on&ci=trunk&name=src%2Fpager.h and another 2,000 implementing a page cache. It’s hard to blame people for making one call to mmap when the alternative is maintaining an additional 10kLoC to do it properly! .

The solution here is basically one of progress. CISA and the FBI have asked vendors to provide memory-safety roadmaps by 2026; while it’s possible to improve the memory safety of C code, these solutions are not complete⁶. Languages built with memory safety by default tend to include syntax that make errors harder to ignore than to handle and include standard libraries with more ergonomic (and opinionated) abstractions for things like file I/O.

That’s 0.5% of the 444.4×10⁹ m³ consumed by the United States alone in 2020², but people need food and don’t need, as one person put it, “bullshit factories”. Which is true! I don’t need “spicy autocomplete” to write code any more than I need a router to cut a rabbet, but both are useful and neither are going away. To continue with the metaphors: these new power tools are themselves in their steam power era.

Steam power is a nice label that helps historians³ avoid having to mention coal. The initial mechanization of all industries (including carpentry) came at such a high environmental cost that “a black form of the peppered moth rapidly took over in industrial parts of the UK during the 1800s, as soot blackened the tree trunks and walls of its habitat”⁴.

The early days of the Industrial Revolution were grim, but there’s not much coal left in today’s industrial power mix—the UK is even on track to shut down their last plant this year⁵. This has mostly been made possible by a combination of electrification and improved grid composition, and the impact has been significant. In 2020, 70.6% of domestically-produced steel in the United States was made by electric arc furnace⁶ and 59% of California’s electricity came from carbon-free sources⁷.

Likewise there are already improvements happening today that directly map to modern datacenters. Within the past decade many municipal utilities have added high quality recycled water (or “purple” water) to their offerings, and renewable sources in Europe now regularly cause power production to exceed demand, resulting in negative electricity prices⁸.

There is cause for optimism, and it’s important that we let ourselves feel that. If anything, it’s inspiring that people’s access to information today has improved so much that the costs of datacenters—which do not directly affect most people!—are such common knowledge; it only adds to the incentives pressuring them to improve.