I don't think Econ 101 price/quantity equilibrium is entirely what's going on here. Gigabit service *availability* is about the same in Spain and the US, despite America's per capita purchasing power adjusted GDP being about 60% higher than Spain's.

I think the relevant figure is this: Spain has roughly 2.8x the population density of the US. It's surely a lot more expensive to build the infrastructure to cover roughly the same percent of the population here.

Although that's not *nothing*, the question is who exercises admnistrative control of that data. If the Chinese government demands data from ByteDance's management, and ByteDance's management complies, that data is not safe. Of course, even in the US a federal agency can obtain a secret warrant which enables them to help themselves to your private data held by a third party, and because it's *secret* you can't challenge the warrant's legality.

The smart thing is don't put anything sensitive onto any kind of social media. Now some metadata may itself by sensitive for certain persons, like your approximate location at various times. Such persons shouldn't use social media at all, even if the data is hosted in the EU, which generally has the best data privacy protections in the world, because there is *no* country in the world where a company can defy a lawful warrant, whatever "lawful" means in that country.

I think one place to expect operational savings is refueling. Conventional reactors spend about 8% of the time offline being refueled. Every eighteen months thousands of workers from all around the country come to the site to do the work. SMRs are designed to need refueling much less often, typical every 3-7 years. Some designs go for up to thirty years without refueling. Plants with a larger number of smaller reactors can also do maintenance and refueling without losing any revenue, as the remaining reactors put out a little more power to compensate.

SMRs shut down and cool down much faster; some don't require any active cooling measures at all. You just shut the thing down and a week later it's cold even if you don't have outside power. So there's a lot less plumbing to monitor and maintain.

Of course these are all just promises now. Running these things is going to be so different we won't really know until we've built and operated some.

"If we only did the right thing then we couldn't do what we want to do."

Which in itself says nothing whether you are or are not violating the creators' rights.

You as the non-owner of the IP have certain fair use rights that depend, not on the mechanism by which you obtain a copy of the data, but on the effect of what you are doing with the data upon the copyright holder's proprietary interests. A download button does *not* indicate content is free game for commercial use.

It's already extremely safe. Cycling has a lower death rate per participant than *tennis*. And while your risk per *mile* is signifiantly higher on a bike than as a passenger in a car, your risk per *hour* is signifiantly lower. Since most cyclists aren't putting nearly as many miles on their bike per week as their car, the bike represents a low risk to them; in fact if you take up cycling your chance of dying in the next year goes down by 1/3 once the fitness benefits kick in, even though you've just added a new way to die to your personal list.

So tech like this is unlikely to reduce *absolute* risk very much, because absolute risk is already very low. It so happens this particular tech could reduce the most fatal type of accident -- being struck by an overtaking motorists -- but these types of accidents are very rare, as are cycling fatalities. Since there's only about eight hundred cyclist mortalities / year in the US there's not a lot of room for improvement, especially as this tech is bound to be installed on only a tiny minority of bikes. It does nothing for the two most common types of accidents: (1) cars entering the street to make a turn and hitting a cyclist traveling along that street and (2) cars passing a cyclist and making a right turn at an intersection across the cyclist's path (the "right hook"), so it's unlikely to affect metrics like ER visits and hospitalizations very much.

We have to sharpen our thinking about what we're actually trying to accomplish when we talk about "making cyling safer". I'd suggest there's two things we can be reasonably trying to do: eliminate as many *preventable* deaths and injuries as possible and make people *feel* safer when riding a bike. There's a lot of injuries that can be taken off the table by designing and marking intersetions better and improving lines of sight. Many of these changes would also reduce car-pedestrian accidents and car-car accidents too.

Technologies like this can't make cycling statistically much safer than it aready is. But they can do a lot to make cyclists feel safer -- much the way some cyclists are spending hundreds of dollars *today* on rear-facing radar units. Those are good things, but they're no substitute for better design which would both make cyclists feel safer and make everyone statistically safer.

Sure, if you just fixed people, that would work better than any conceivable technical solution.

But you can't fix people, so there's no point in complaining that *some* people are *sometimes* stupid, careless, or irresponsible..That will never change. Sometimes, even, that careless person might be *you* on a bad day. We all rate ourselves based on our performance on our good days; which is why we all think we're better-than-average drivers, but really our risk to ourselves and others is dominated by the days we didn't get enough sleep, are stressed out, distracted, and running late. Those are the days when the things we habitually do right go out the window.

If you frame the root of the problem as being "people aren't good enough to operate the world we've built", then you're stuck putting band-aids on the problem. That's not unreasonable as it sounds, think of how things would be if you didn't *have* band-aids. People would be getting limbs amputated because otherwise insignificant cuts got infected.

So it's perhaps better to frame the problem this way: the world we've built is not suitable for people to operate in with acceptable safety. The root of the problem is design, so that should be highest priority. But even so, we'll still need to address the failures that better road design can't fix. Even the most safely run factory still needs a first aid kit, and that first aid kit is going to be stocked with band-aids.

But it matters to the research and operation teams whether their downtime is two months or nine months. Such a reduction could alter the economics of a robotic exploration program, which would surely be a prelude to a manned mission. So the robotic program could both provide input into planning of the manned mission while proving the propulsion technology is reliable enough to be man-rated.

Current concepts for a manned Mars missions would last 22 months, 21 of which would be spent in transit and one on Mars. If the round trip transit time were reduced to 4 months, you could spend 18 months on the surface in a mission of the same length. In such as scenario with robotic missions you could avoid staging supplies that only *might* be needed knowing you could send them if the need arose *during* the mission. You could respond to unexpected circumstances, or return samples to Earth for analysis that could alter the priorities of the mission.

So it could make sense to build unmanned vehicles with such a technology -- as part of a manned *program*.

Peer reviewers are volunteers who don't get paid, at least that's not the norm in science. Arguably they should be given the importance of the task. If you've ever seen peer review comments, some of them are obviously phoned in. Occaisionally institutions will offer honoraria for reviewing proposals -- typically $200 or so. This is not a lot of money considering how much work it is.

Here's how I look at it: generative AI is dsigned to create plausible-looking output in response to a prompt. That's how the lawyer who submitted a ChagGPT-generated brief got caught. The references the AI generated for the brief looked plausible enough to pass a cursory inspection, even by an expert, but if you looked them up they didn't exist.

This wasn't a failure of the AI; the AI did exactly what it was designed to do. It was the fault of people who relied on it to do something it wasn't designed to do. Surely someday soon a *lot* of the work of peer review will be automated by AI before a paper actually gets reviewed by a human, to avoid wasting reviewer time with obvious shortcomings.

Systematic reviews, including meta-analyses, are very common. Since roughly the 1980s systematic reviews have been considered the highest possible tier of scientific evidence.

This is because of certain facts about science that outsiders often find shocking: (1) complex questions nearly always have contradictory evidence and papers taking opposing views of issues, particularly early on; (2) every paper, no matter how good, has methodological shortcomings if not outright errors; and (3) many novel findings never get replicated. This means individual papers are almost useless for proving anything, at least without putting them in context.

That's what systematic reviews and meta-analyses do: they put individual studies in context of what other researchers are finding. Unlike some internet rando citing papers to prove his pet theory, a reviewer can't cherry pick papers based on what they find. There are rules that ensure papers can only be excluded for objective reasons that apply to all the papers on the topic.

In this particular paper, the authors did not find evidence that deforestation and forestation were drivers of disease. An activist writing a polemic wouldn't come to that conclusion.

The company management is pointing the finger at workers, and they're right to, just as long as they point the finger at themselves too.

These kinds of problems start at the top. If management demands workers do the impossible (or at least the wildly implausible), they know that reports of success are going to be fraudulent. The question is, are they goign to get away with it?

I really think the main argument *for* carbon offsets is that it *potentially* can harness free market mechamism to *efficiently* reduce emissions. This would be in contrast to a pure government mandate that everyone cut their emissions by some percent. The problem is that the marginal costs for industry X might be prohibitive; on the other hand industry Y could easily cut more. So why not have X pay Y to cut more than required? This *internalizes* the external benefits of extra reductions for Y.

Of course, it's very easy to screw this up, starting with letting people get away with fraud. But if you allow fraud in *any* market, that undermines the efficiency of the market. If you are going to get the entire economy to reduce emissions by some set goal, you need some mechanism to distribute those reductions so they're made where it's most efficient, and financial efficiency is one thing the free market excels at.

So let me get this straight. If there were *real* carbon offset credits, that would in your view be *just as dishonest* as fraudulent ones?

how?