xablor
1yr ago
"Interactive Theorem Proving and Program Development: Coq'art: The Calculus of Inductive Constructions", and it's kicking my ass to a humbling degree. I'm spending, conservatively, 15 minutes per page, in chapter /one/. I don't know if I'm dumber since I was in undergrad, or if this is my true info onboarding rate and I did undergrad wrong, or what, but this isn't boosting my ego at frigging all.
xablor
1yr ago
Ethics question: how evil would it be to develop a payload for a mechanically suitable off-the-shelf remote-control multirotor drone that would enable a remote user to pierce a car or truck tire and render it irreparably leaky?
For numbers, let's say:
-
the drone is viably controllable up to a quarter-mile from an off-the-shelf controller station (read: phone or lap, maybe with a radio dongle)
-
the drone is not autonomous outside basic flight stability and safety features to other humans, so it has to be guided to a tire and the knife triggered by the user
-
the knife can be triggered 4 times per flight
-
the drone's battery and knife can be replenished within a minute by the user
-
the knife is captive, so it can't hurt anything the drone isn't immediately adjacent to, and magically can't be modified to do otherwise by end users.
-
the drone and ground station are readily replaceable for <$10K, so accessible for a small organization or an org with donors, but not a typical individual.
This is prompted by my trying to inhabit the viewpoint of modern dirtbag left activists, such as those who protest by gluing themselves to roads and suchlike.
Factors I can think of offhand:
-
This enables grassroots enforcement of no-car, no-truck zones for the anarchistically-inclined
-
This makes destruction of property safer for the perpetrator
-
This enables wider-scale destruction of property viable for a single user
-
The payload designer isn't hard to replace, since the payload is easy to design, but the payload only needs to be designed once and then plans distributed
-
Obviously, this makes hit-and-run violence easier and safer, but that rate is already low and dropping, but maybe someone out there is only held back from a spree by having to be present for the attacks in person? If so, why aren't they a sniper on a spree already?
-
Once the payload is built, how much harder is making the entire thing autonomous? To the degree of "here's a car-shaped thing, slice the tires"? "Here's a geofenced area, slice the tires of all car-shaped things in it"? "Here's a geofenced area, slice the tires of all cars without a badge"?
xablor
1yr ago
That was my thought, yeah. Some style of Derringer-style action with a captive spring-loaded blade or piston, which is what drove the small handful of pierced tires between visiting a human.
xablor
1yr ago
Is there a pinned thread somewhere for feature ideas/requests/feedback? My Motte-fu is weak.
xablor
1yr ago
Is this a game for people with compromised shoulders? If I can't do an overhead press without subluxing, will driving stress that failure point again?
I caught this exchange after the previous thread had mostly closed, and I'd like to push back on the claim a little.
BinaryHobo:
I remember talk about just using the excess power to pump water up hill during the day and running it through turbines coming down at night.
Did anything ever come of that?
The_Nybbler:
The physical conditions necessary to make hydro storage practical aren't common.
(How do we do the fancy quotes with user, timestamp, and maybe a link? It'd be useful here.)
It's true that hydroelectric power sources, as in dams, have saturated the supply of naturally-occurring American sites. You need a river in a rocky valley, and there are only so many of those to go around, and once they're used up, it's very hard to create more of them.
What haven't been exhausted, and in fact what can be readily found or exploited, are height differentials in general. Hills, mountains, exhausted mines, deep valleys with no water supply, all offer significant height differentials, are naturally occurring, and can be readily built out into large-scale closed-loop pumped-hydro storage, with a closed reservoir at one extreme and a closed reservoir at the other, and a reversible turbine to generate potential energy in times of excess and power in times of deficit. Should those be exhausted, off-shore dropoffs are an enormous resource of the same, at the cost of more difficult installation and operation in every regard. And if we exhaust THOSE, water towers at sea or underground reservoirs on land can be constructed as well.
All of this, of course, is dumb and America should just take the leash off nuclear, as argued here. (I've not read it yet, but I expect it to make the points I would inline here.) That we haven't yet is a shame and a testament to our collective idiocy and Puritan hangover.
xablor
1yr ago
This one is extremely impractical, which you’d see if you even did a back of a napkin estimate.
Source. They've tested successfully, physically, to 1/10 scale. I haven't gone and found the paper, I'll admit; I'll give it a shot ASAP so we can argue productively.
In the meantime, if the napkin math is so easy, share it with the class?
xablor
1yr ago
I haven't gone and found the paper, I'll admit; I'll give it a shot ASAP so we can argue productively.
I've casually found:
xablor
1yr ago
The goal here is storing energy for use later at a net loss, not harvesting energy for "free" like a dammed river reservoir that sometimes gets rained on. Inefficiency is acceptable, although to be minimized.
xablor
1yr ago
An excellent summary is given here .
Tldr, largely copypasta:
-
The American Nuclear Regulatory Commission uses a model of damage to humans by radiation called Linear No Threshold, in which no amount of exposure to radiation is safe. This contradicts casual observation (we live with and robustly tolerate background radiation), observed cellular mechanisms (detection and repair of small DNA errors is routine), and a small number of human longitudinal studies and animal studies.
-
American nuclear reactor operators are as a consequence required to minimize the risk of even innocuous, low-level radiation releases, which makes cost reductions as a result of the usual learning curve and technological advancement impossible.
-
Culturally, there is little education on the risks of small and medium-scale nuclear incidents, and so public opinion is by default against radiation leaks out of proportion to the actual risk. The book being summarized contrasts this with airline accidents, which kill hundreds and are handled as a risk to be minimized, not eliminated.
-
The NRC is incentivized to run the approvals process as long as possible, since it collects fees from license applicants, rather than number of nuclear power plants under oversight or number of GW-hrs generated by nuclear power per year. This naturally drives up the costs of site licensing and design approvals.
-
There are many avenues for anti-nuclear activists to cause delays in the construction of a nuclear power plant, causing massive uncertainty in construction schedules and costs.
-
A model reactor must be licensed before construction begins, but model reactors are often invaluable in experimentally finding failure modes to be accommodated, but all possible failure modes must be addressed before even a model reactor is approved for construction.
-
Chernobyl, Three Mile Island, Fukushima incidents have accumulated massive cultural scar tissue opposing more nuclear power plant construction.
Tldr of tldr: ignorant public, regulatory incentives, uncertainty in capex and opex spend.
xablor
1yr ago
I'm a technologist, I'll propose a tech solution off the top of my head: containers for digital media (video, audio, hell why not text documents or tweets) that's trusted for public decisions must be cryptographically signed and checksummed by the originating device before it hits userland, and further signed by a trusted location service that claims the capturing device was actually present. Media without the container is considered not possibly trustable. Unpacking the container and doing ANYTHING to the contents, without the private key of the originating device, becomes detectable. The entire problem reduces to a) key management, which is merely moderately hard at scale, but made easier by centralized management for many (almost all?) mobile device, b) the trusted location service, c) protection against extracting the signing key of the originating device. Obviously this trades off significant amounts of privacy, but if you're submitting your film to the MSM to influence public opinion, that's maybe acceptable.
Taken together, this should push the cost to get media trusted by the MSM to state-level attackers.
xablor
1yr ago
I bet the Naltrexone is prescribed daily? You might consider using it per the Sinclair method (protocol when you're defending it to your doc), where you dose an hour before drinking. This blocks the reward from drinking more intensely and more precisely than once-a-day dosing.
xablor
8mo ago
Inspired by college loans discussion earlier, I'd like to apply a policy engineering lens:
- What are the minimal changes necessary to get an epsilon away from the current model of "government guarantees the entire loan amount with no conditions and it's not dischargeable in bankruptcy, with private lenders available" and towards a model that incentivizes better behavior in schools and students?
- Politically, what gets the nose under the tent most effectively, allowing further reforms?
- More generally, it seems difficult to implement a series of small reversible reforms to explore a space; what drives that in a government implemented as largely autonomous opaque agencies, and is it itself reformable?
xablor
8mo ago
Can we add an iterator over new comments since the last time a user loaded a thread? They have their own unique CSS class, looks like, so the server-side logic already exists to detect them. This might even be doable purely as a client-side bit of JS.
xablor
8mo ago
Off-the-cuff gradualist proposal to play with, and maybe be an existence proof: fedgov only guarantees 95% of a loan balance, measured as an amount of approved loan principle. Or whatever amount is needed to see a signal emerge (is part of your point that the minimum amount to see that is too large to be viable?). This paired with, I dunno, earmarks for grants for some of those who are impacted, to add palatability to lefty opposition and keep incentives straight for borrowers.
What doesn't work there?
xablor
8mo ago
Naive policy engineering again, American electoral reform edition:
Team Red claims to want "reinforced" elections, where the risk of people casting a vote who shouldn't be able to is minimized or eliminated. A common proposed mechanism is to use state IDs to validate that the holder has the right to vote in that state or federal election, and (I imagine) to enforce one-vote-per-person. They prefer the decision to be biased in favor of minimizing false positives at the cost of increased false negatives and possibly true positives.
Team Blue opposes this with rhetoric about wanting to maximize access to the electoral systems at all levels. They prefer to maximize true positives and minimize false negatives at the cost of false positives, the symmetric opposite of Red, as in all things.
Left unstated is the assumption, seemingly held in common by both Red and Blue, that people who have a hard time obtaining state IDs are likely to vote Blue.
A compromise solution seems to exist, and I don't understand why it's not being pursued: increase funding for voting accessibility programs, in exchange for tighter requirements for voting authorization. Have, literally, a list of people who were born in state, can't be accounted for as having left the state, and authorize a spend of $10k or whatever to find them and Get Them Registered No Matter The Cost.
One thought: spending on this is a continuous value, whereas a policy state IDs as a bearer authentication token are boolean. Fine, hold state IDs out as a carrot, and offer improvements in, I don't know, signature matching in mail-in ballots.
In summary, two symmetrical problems exist, there exist opportunities to progress towards solving both of them, no serious efforts are being taken. Why? Per the meme, are they just stupid?
xablor
8mo ago
Because ‘voting accessibility programs’ are just funding for democrat political machines
I mean. Yes? That's the point, to give both sides part of what they want. Democrats want funding for democrat political machines.
tighter requirements for voter authorization are totally off the table for democrats
This I'm more ignorant on. There really aren't any minority sub-parties that don't care about the topic that can't be peeled off?
xablor
8mo ago
What compromise could the right offer the left that they would want?
Did you actually read the (offhanded, I admit) proposal? It was short. It involved creating a state-level capacity to locate disenfranchised (read: poor, illegal, disabled, low-executive) people and extend state services to them. Inside five years it'll be the premier way to access lumpenproles to buy votes from them. What compromise is possibly better than actually, actively, implementing your opposition's agenda for them?
EDIT: You reacted to it, so you must have. I am confused.
xablor
7mo ago
You might be encouraged by recent work in current-gen small modular reactors. Oak Ridge National Labs is standing up 6 of them, a few efforts got approved to start construction, and the US military is obviously interested given their focus on expeditionary logistics.
This is very cool forum site, kudos to Zorba et al for their hard work and a smoothish launch.
In the vein of navigating through read comments to new ones, it would be cool to be able to see how many unread comments a collapsed comment has beneath it. Maybe this could be pushed to user-side logic? How does the unread-comment counter work for the thread index page?
More options
Context Copy link