- I care about injuries other than just fatalities
- Non-fatal injuries due to being hit by cyclists are underreported
- Those statistics are for all of NYC - the pedestrian experience in much of the Bronx, Brooklyn and Queens, and all of Staten Island is similar to the pedestrian experience in suburbs - and I'd agree there my danger is greater from cars.
- Those statistics pre-date the dramatic rise of e-bikes on the streets of Manhattan
There are research on how many accidents are reported and at least in Sweden most accidents that require hospital visits are reported correctly. (and they mimick what is said in other countries).
Streets with bicycle paths are a safer for pedestrians, I think there is research from NYC/Brooklyn comparing streets before and after bike paths.
Ironically the Sierra Club is one of those organizations behind NIMBY type objections to solar and wind production. They are an outdoorsman club, not an environmental club. Their objective is maximizing the amount of untouched wilderness available, and that means not being able to see wind turbines from mountaintops.
OP describes systems which seed tests with non-determinism to catch users who write tests relying on undefined behavior, such as order of equal elements. Writing tests against these systems is a game changer (once you figure out what’s going on and why there are flakes). I’m a huge fan of this subtle change which has spared me at least one bug. It also allows library authors to gain the advantage of not accidentally producing a behavior that induces Hyrum’s law. Kudos to the team for making this work at scale.
Not mentioned in the article is power density. How quickly can the energy be released? Consider solar panels, you need a table sized cell to get 100W. That can make for a big battery to get grid scale power output if these cells are only as power dense as solar panels. The energy density of a heat based solution can be very high- metals can get very hot and they are dense enough to store a lot of energy. But if you can’t get the energy out of the battery fast enough that limits the applications. By comparison lithium ion batteries can dump power out extremely quickly, which is what makes them great for cars. Hydro is even better.
The article in Nature quotes an energy density of 2.38 w/cm^2. Which means a Gw battery would require 10e5 m^2 of absorber surface, exposed directly and at close range to the radiation from molten metal (which is the heat transfer fluid they propose). It has to be direct, and at close range, because the efficiency they quote relies on the absorber reflecting non-absorbed photons directly back into the emitter, where they are re-absorbed as heat and potentially re-emitted.
That's about 25 acres of absorber, and an implied 25 acre surface area of the liquid metal emitter pool.
There is a basic challenge here to the design - the energy storage density for the thermal battery they envision scales as the cube of the characteristic dimension of the plant, but the power density that can be delivered scales only as the square of dimension. Not saying that can't be dealt with in engineering, but it ain't going to make this easier or cheaper.
Surface area is relevant for solar because the sun is so far away. A local heat source allows you to surround it with 3D shapes not just a flat plain.
As to temperature this thing is for very high temperatures: can generate electricity from a heat source of between 1,900 to 2,400 degrees Celsius. At 40% efficient you need a wide temperature difference which would suggest a high energy density.
This design is photovoltaics, just like solar, but optimized for infrared photons. There is no avoiding the reality that energy storage density will scale as the cube of the facility size, but power density only as the square. And at 2.38 w/cm^2, the scale coefficient is not all that great.
Picture a stack of flat panels with each layer consisting of: (Cold)(Panel)(Hot)(Panel)(Cold) held vertically. So: (Cold)(Panel)(Hot)(Panel)(Cold)(Panel)(Hot)(Panel)(Cold)(Panel)(Hot)
Now you add hot gas at the bottom and have say 4 layers per m. So a 3mx3mx3m cube would be 4 layers * 2 panels per layer * 3m * 3m * 3m = 216m2 of panels taking up a 3mx3m section of floor. At 2.38w/cm2 * 10000cm2/m2 * 216m2 = 5.14 MW of power.
For long term energy banking and if we can get them working, flow batteries seem vastly superior to all alternatives, by scaling storage with regards to tank volume. Instead of some difficult-to-manufacture structure.
I think their application is grid scale and you can scale across hundreds of batteries to provide the throughput you need. I don't know how I feel about having a small molten ball of metal inside the hood of my car. Turns my car into the most dangerous gusher in the case of an accident (for those who aren't familiar, gushers are a gummy like candy with juice inside).
I care less about hooking into range and more about the performance. I would really like for iterators to be zero-cost abstractions, but right now a SliceIter is waaaayyy slower than looping over a slice (with range or otherwise). The Go compiler isn’t smart enough to realize that a loop over a SliceIter can be reduced to a loop over a slice, perhaps because it can’t know for certain that the loop counter variable (e.g. the `counter` field in `type SliceIter[T any] struct { slice []T; counter int }`) is only being accessed by the loop itself (this might be where Rust’s ownership semantics would be useful in Go?).
I was curious to see if anyone had tried the coal mine bit. This is the closest I could find, a business plan to sell coal attached to carbon credits. Not sure what’s happened since 2016.
Greenpeace had claimed that they tried to offer Vattenfall to buy their coal mines in Eastern Germany to shut them down (they were later sold to czeck company EPH), but it's a bit unclear how serious that offer was:
https://www.fr.de/wirtschaft/greenpeace-bietet-kohle-tagebau...
Though maybe that's also a cautionary tale. Back when that happened (2015) the political discussion around coal in Germany was that a coal exit wasn't really part of the discussion and local politicians would say things like coal will be mined until the 2040s. Since then the debate has moved to "Coal exit in 2038", and more recently to "Coal exit very likely in 2030".
What I'm trying to say: Even if you're successful with such a project, it's not exactly clear how much you gain. The political discourse on climate is moving fast (though still not fast enough).
I worried about this too. But I think my attitude has changed after watching lots of the PBS Space Time youtube channel. They do a great job of breaking down these concepts at a level where highly interested non-physicists can get what feel like the real details without dumbing it down too much. They have good videos on many physics topics, and regularly explain new discoveries.
The Fermilab channel is also quite good for short-form content. ScienceClic English has absolutely wonderful visualizations. All of them do make some subtle inaccuracies or skip things for the sake of brevity though. I think Sabine Hossenfelder's channel has the most accurate videos in that 10-15 minute range, but they're still only about 15 minutes long.
I don't have a proper education in Physics, but have been trying to self-teach and I think that none of the ~15 minute video channels really cover things to a very detailed degree. You really do need textbooks/lectures/real papers to actually understand it. The channel "Physics Explained" is pretty good for more in depth breakdowns of things, but it is quite dry compared to those other channels and still not really a substitute for a textbook or class.
And I don't even mean learning things well enough to get a job as a particle physicist or anything. Just some things, like say particle spin, just can't be explained in under a few hours and without the math behind them. They don't have a proper intuitive analog to our macro-level world.
PBS Space Time seconded. I recommend taking the rabbit-hole approach with them - i.e., blocking out a considerable more amount of time than the length of the video you're about to watch. They always reference past videos that expand on the building blocks of whatever the topic is in the video you're watching, and it helps to go watch those if it's a new topic to you, before continuing with the current video. I absolutely love that channel.
Thirding PBS Space Time, and seconding taking time to really focus and treat watching them like studying. I became really interested in physics about 2 years ago. Initially the content was really challenging but I forced myself to rewatch many of the vids several times and it was ultimately very rewarding. I have a good enough grasp on the core concepts that I'm able to explain them to friends in-depth and it makes for great conversations, especially when people are in a state of mind to pontificate about the nature of the universe hehe.
Fourthing as well. Think I've watched (heard) every episode at least twice now. They really calm the mind I find, as I frequently drift off to sleep with it in the background, dreaming about Space Time.
I like Sabine's channel even better. She is great at simplifying things, and explaining the raw concepts and what some equations / findings really mean
If a Netflix subcontractor didn’t pay their actors, writers, etc, they would catch a well deserved deluge of criticism. Of course it wouldn’t happen because those workers are organized to prevent these abuses.
I don’t think this is cut and dry, and instead depends on the nature of the relationship. To buy a laptop, Netflix didn’t have board meetings, deep contract negotiations involving the executives of both businesses, or an ongoing relationship.
The laptop is effectively a commodity, and any injustice Apple commits is amortized over many million laptop owners who tacitly but minorly endorsed that injustice. Netflix buying a show feels like a much more intimate relationship where much more of the responsibility lies with Netflix as the only buyer of the product.
Netflix isn't the only buyer though. They make the show and sell it to whichever studio compensates them the best. And then sometimes sell it to a different studio in a different country.
Show purchases don't go past the board. They don't even always go past the CEO.
It's really more akin to buying enterprise software. So sure, change the metaphor to Oracle database. You don't complain about Oracle's customers where Oracle does something bad.
That's a bad metaphor, though, it's more like if Netflix hired Apple to make a special type of Laptop that only Netflix will sell, and the laptop will be marketed as "The Netflix Laptop".
I haven’t done the analysis myself, but I highly doubt winterization would double the cost of energy. It’s probably some small portion of the capex to build the plant in the first place. Would you pay an additional 5% or 10%? Probably, especially if you get to avoid your home being flooded by burst pipes, rendering it potentially unlivable for much longer than the week of power outage.
You'd be amazed what can be done with the latest machine learning algorithms. But setting those aside, you could still get 90% of the way with some relatively basic stuff.
I'd start by identifying the presence of speech—when the user isn't currently speaking, trivial muting is all that's needed. Then the only challenge would be to squelch keyboard noise that overlaps speech. If keyboard input monitoring has enough temporal precision, even the most trivial volume dip (and perhaps a band pass filter) would be a win.
Keyboard input monitoring could also be used to continuously tune the algorithm to the loudness of the keyboard noise. Any key press occurring in sonic isolation could be used. Then when there's keyboard noise which overlaps speech, the algorithm will know how much to squelch.
https://www1.nyc.gov/html/dot/downloads/pdf/nycdot-pedestria...