@blooblyblobl's banner p

blooblyblobl

Battery-powered!

0 followers   follows 0 users  
joined 2022 September 04 22:46:30 UTC

				

User ID: 232

blooblyblobl

Battery-powered!

0 followers   follows 0 users   joined 2022 September 04 22:46:30 UTC

					

No bio...


					

User ID: 232

The UAW is still going strong... Clearly someone has a use for low-skill, high priced, lazy, dishonest labor. Just maybe not the people required to hire them.

In any case, I think "lack of English language comprehension" is the real nail in the coffin. We can get away with Spanish since it's basically the unofficial second language of the USA, but I don't see too many government forms translated into Creole.

The fiber I mentioned has been replaced with newer, higher-bandwidth stuff a couple times in the last 40 years. The key value proposition wasn't the fiber itself - it was laying a channel with enough access ports that anyone could run whatever they want through the run for the next hundred years. In this case, the power company paid for the fiber itself, then businesses paid for updates and replacements decades later, and it cost a fraction of what it would have to tear up all the streets and put in new channels everywhere. Like @FiveHourMarathon mentioned in the sibling post, leaving room for expansion and future updates is smart.

Greenfielding new commercial construction? Absolutely, put it in now. Back in the 80s, a family member's residential construction business put in a fiber network for the power company in a front range city, and the one-time installation cost has paid for itself about a thousand times over, even allowing for updated runs and municipal gigabit fiber to the home in that neighborhood. Smart infrastructure investment is usually a good deal, even if it's pricey up front.

Just don't make me re-wire my entire panel box for solar/battery/EV deployment, when I have none of those things, just to finish my damn basement.

At a county level, non-mandatory code compliance can once again become mandatory. My father runs (well, ran) a business remodeling homes in and around a city where such adjacency to reality is the default disposition, and the cost and time difference between projects in this county and projects in the neighboring farmland county would make your head spin. Pulling permits for just about anything is a nightmare, since the city takes this as an opportunity to force homeowners to update their properties for modern code compliance. We've had deck jobs where the city-mandated updates to insulation, electrical, etc ended up costing more than the damn deck.

The city council loves bragging about how they have some of the strictest and most complex building codes in the nation, because in their minds, it means they're doing the most in the fight against climate dragons. It somehow goes unmentioned that a lot of these initiatives are counterproductive or dangerous (mandated over-building electrical infrastructure for any job that touches an electrical installation encourages people to get creative with their own wiring jobs). A lot of the other mandates are clearly back-scratching graft for a captured industry - the insulation requirements are so ridiculously overprovisioned that in many cases the only way to meet the county requirements is with special sealant foam along every wall joist, which costs like $3/ft and is sold by one company in the state. Note that this can also be an add-on requirement by the county for anything that opens an exterior wall for any reason (including maintenance and repair). This all has exactly the expected effect on willingness to perform maintenance by local landlords, and large swaths of the city have properties full of dozens of trivial issues that get swept under the rug to avoid incurring massive update costs unrelated to basic maintenance. And of course, the dilapidated ghetto-houses in this city are worth double or triple the value of the neighboring county's much-nicer modern homes, contributing another unironic brag by the city council.

During the pandemic, out of an "abundance of caution" the city council decided to shut down the building and planning department - offices closed, phones off, no one responded to emails - and moved all the permitting forms online. The online form was totally broken, of course, and auto-rejected all applications. For 18 months, you simply could not pull a permit in the county. Homeowners were understandably upset with this state of affairs, and eventually ignored the permits and moved ahead with their projects. As long as we follow code, and get engineering approval for anything that needs it, the only thing missing is the formality of a city stamp on your piece of paper, and they're not supposed to arbitrarily reject you if you're doing everything right, so... Of course, the instant the city comes back from their 18-month paid sabbatical, they immediately start suing every homeowner who did unpermitted work, starting, of course, with the people who live in viewing distance of the city council members! The city council got a reality check at the state court level, but not before wasting millions of dollars attempting to punish their own citizens for a problem they created.

Most of the city council secured re-election.

My dad has since left the state.

I knew before I previewed the link that this would be the ciechanowski animation, and I'm here to express my extreme delight with everything on that site. The GPS animations are also stand-out spectacular. One of the few people on the internet doing interesting things with the new WASM toys.

Maybe! I like to forget which is which. It makes for some funny threads.

I met a call for indiscriminate mass murder with a self-regulating incentive system that simultaneously brings out the best in people, offers a second chance to those truly down on their luck, and condones the death of the undeserving - I'd hardly call that "unbounded sympathy".

On a serious note, I totally agree that there has to be a limit to society's generosity for recalcitrant insanity and unrepentant antisocial behaviors. I also think that, as far as solutions, "kill them and everyone that roughly matches that description" is a lazy edgelord hot take; the ridiculous cost of food and shelter lately is probably responsible for a considerable fraction of the "roughly matches that description" class; and there's an important distinction between criminal and personal nuisances.

Bailey Not-Castle: let's make homelessness punishable by death!

Motte Castle: let's kill everyone on the "individuals known to authority" list, as they commit the Pareto majority of homeless nuisances.

I won't poke too much more fun at this, since I did literally ask for it, and clearly connecting punishment to crimes instead of statuses is a promising step forward.

I can at least respect this position. Taxpayers and charities have handed lots of money and time to various entities to fix the problem, and they have a nasty habit of either making the problem worse, or running up a huge bill to sit around and pontificate on the problem. The police are neutered, incompetent, apathetic, or incapable of dealing with the problem, often by the demands and threats of a tiny slice of the activist class. And the homelessness problem has visibly gotten terrible! I live somewhere where I've seen firsthand how bad things have gotten. I can understand why people are eventually going to reach for vigilantism or mob rule when every function in society designed to protect against these problems has failed or turned traitor to wage class warfare.

And if we reach the point where our leaders, police, activists, and technocrats really can't fix the homelessness problem, and the only solution really feels like mass murder... indiscriminately taking out our collective anger on the mentally ill, addicted, and financially unlucky, is missing the forest for the trees, no?

Oops, thanks.

A modest proposal, then?

At the risk of taking the bait, and against my better judgment... This is a hideously lazy solution for a society that has moved beyond sustenance farming. You're making a cynical presumption of intentional apathy to justify unreasonable measures, when in reality, there but for the grace of God do you go. In a society where your friends and family reflect your attitude, you're one TBI away from being labeled an inconvenience and put to death. Without a hint of introspection or irony, you condemn the homeless because your tiny slice of the collective burden of housing them is too costly and inconvenient for you? Have we considered, perhaps, making the burden less burdensome? Maybe eliminating legislative barriers to affordable housing erected by the economically privileged would be a better place to start than getting out your guns and going postal on a tent city? As it stands, you want society to grant you a license to kill those who inconvenience you - and this is exactly the sort of small-minded, impulsive criminality that society is constructed to curtail.

While we're unseriously venting our spleens, here's a modest counter-proposal: you can have your license to kill the homeless, but you only get to kill as many people as for whom you voluntarily provide housing, and if you ever stop providing that housing for any reason, and anyone you housed is killed by this policy, you too are put to death. This is at least marginally less lazy than your proposal, because it forces you to exercise discriminating judgment as to who is worth helping and who is a lost cause, and it guarantees that you can't take someone in for a day just to execute them the following day. You get to slake your bloodthirst and prove that you aren't just a lazy sociopath who wants society to give you a free pass for murder; I get you to rehabilitate or hospice someone who doesn't deserve to die, because your skin is in the game; some people get a better life than they currently have; and we get to eliminate the truly hopeless cases. And if no one agrees to house someone for their license to kill, we're no worse-off than we started.

I'd personally prefer if we don't openly advocate for killing groups of people over inconveniences and unproductivity, but if we must, let's at least try to address the obvious, foreseeable objections to our modest proposals with our own well-reasoned conclusions, and not just show our whole ass to the world?

Hey now, there's definitely a third option.

If I were Biden, I'd keep an eye out for stray banana peels.

I realize now that I should have been saying de facto legalize. Yes, bribery is still illegal, and it can still be prosecuted, but it's very hard to go anywhere with that prosecution if the President can declare the initial act of discussing a bribe as part of his authority to seek opinions from his officers, thus rendering it subject to simple absolute immunity, and preclude the court from considering it (including for the briber, for what it's worth). Barrett's objection concerns the quid-pro-quo evidence created by the President after agreeing to a bribe; my concern is with the impossibility of demonstrating the President's (or his briber's) mental state, a necessary prerequisite for a bribery prosecution, when the court is explicitly disallowed to consider it.

The constitution clearly delineates between "Law" and "Impeachment", and the two are unrelated. I see the word "nevertheless" here meaning "this clause is about impeachment, not about the method of criminally prosecuting the president in spite of their absolute immunity", rather than "but if you remove them from office, you can now also do this other stuff".

I could be wrong, and if so I think you make a compelling argument for what right looks like.

Edit: See The_Nybbler's point here.

But again, this does nothing. There's no declarative requirement for the President to e.g. invoke executive privilege. The President has absolute immunity for seeking opinions from his officers, it's not something he has to argue, he just says he was seeking an opinion and that's the end of the discussion; you can't subpoena parties or submit records of the conversation as evidence, and without that, you have no evidence of bribery. And as far as I can tell, impeachment (even after leaving office, which is out on a limb at best) doesn't strip absolute immunity. There is no legal battle, because the required evidence to prove a crime or any circumstances under which immunity wouldn't apply, impeachment or otherwise, cannot be considered by the court.

The outcome of an impeachment is, at worst, removal from office. Nothing about impeachment appears to grant anyone authority to strip absolute immunity conferred to actions taken pursuant to constitutionally granted powers of the President while the President was President.

ETA:

Judgment in Cases of Impeachment shall not extend further than to removal from Office, and disqualification to hold and enjoy any Office of honor, Trust or Profit under the United States: but the Party convicted shall nevertheless be liable and subject to Indictment, Trial, Judgment and Punishment, according to Law.

I'd read that as allowing subsequent prosecution, but not somehow removing absolute immunity from actions taken while still the President. If not, you could plausibly impeach and try every president immediately after they leave office, or whenever an opposing party gets a majority in the Senate, for crimes even intended to be obviated by the president's official powers. In which case, why grant immunity at all?

what's stopping the next president from ordering the DOJ to prosecute and investigate?

Nothing, they just can't build a case if it depends on evidence precluded by simple absolute immunity.

What's stopping Congress from issuing a subpoena and hauling the last guy in?

The aforementioned simple absolute immunity.

What's stopping a federal prosecutor from opening a case and bringing charges?

Aside from bribery cases being guaranteed to fail in the absence of evidence of mental state, nothing.

For bribery specifically, you must prove the President's mental state of knowingly taking a bribe. The majority ruling explicitly forbids the courts from considering the President's mental state when determining if actions taken under the powers granted by the constitution are improper. And the president talking to his executive officers is explicitly a power granted by the constitution. So if the President says he wasn't taking a bribe, he was just talking with his officers... End of story, this crime cannot be prosecuted because evidence to the contrary is subject to absolute immunity, and is required for conviction.

What good does it do to prosecute a case if, by definition, you are guaranteed to fail?

This is susceptible to last-day shenanigans - if Congress can't impeach you fast enough, do you get away with it?

I'm not too concerned about the military implications.

On the other hand: Did Trump v. United States legalize bribing the President?

If the President has absolute immunity for discussions and actions relating to his constitutional prerogative, and bribery requires demonstrating a state of mind, typically through evidence that shows intent to bribe or be bribed, how would the Justice system examine such evidence if the President pinky-swears that it's legitimate, and is therefore subject to absolute immunity?

They do industrial espionage, they're very rich and can pay people who developed these processes for ASML, TSMC and so on more than they would get paid in the West.

In spite of their best efforts, they still have no domestic EUV industry to speak of. Contrast with the domestic advanced semiconductor industry they do have. Clearly one of these is harder to replicate than the other.

They do pay well, provided you don't mind living in a country where the government can and will disappear or execute you for wrongthink, regardless of your station or importance to the technology roadmap. Many such cases!

just built some more nuclear power and bigger datacenters?

Again, no one can stop them with sufficient motivation and willpower. But it can certainly be made to take a while and cost a boatload. As HPC systems get bigger, there are topological and latency challenges that compound on each other. And they will have to keep getting bigger, slower, and less efficient to keep up with the pace of innovation in equivalent western systems. It's not an insurmountable obstacle, but it costs resources and time. ETA: More power also means more cooling required. More power density in less efficient devices places huge demands on cooling, or forces more lateral scaling, which also compounds the problems above.

I'm assuming that the big concern with high-volume Chinese HPC is them achieving AI dominance and outpacing western innovation long-term - correct me if I'm wrong. I think, for very large AI, the software, and particularly the training techniques, have orders of magnitude of improvements awaiting discovery and implementation. This could go either way, western or Chinese advantage, with the understanding that Chinese industrial espionage is highly effective at extracting and implementing mathematical innovations in software, substantially moreso than hardware. I think it's more likely that the high costs of hardware force China to be parsimonious with their compute resources and ultra-efficient with their software craftsmanship, than that they just steamroll through the obstacles and disregard the expense.

Chinese have a higher supply of STEM grads of the appropriate intellectual level than the entire West, so 'catching up' for them is only a matter of will and investment.

And, most critically, time. EUV took decades to get to the state where it could barely be called working. Remember the mythical man-month? China's supply of STEM grads is sufficient to undertake nation-state scientific giga projects, yes, but they will be rediscovering the required physics and engineering for years. And until they figure it out, they're stuck at quad-patterning DUV yields (which aren't great below 14nm), or even worse yields on even more multipatterning. And that means more subsidies the government has to pay to top chip manufacturers to produce otherwise unprofitable chips. I know you linked to a guy who says they got the yields up above 50% and they can probably improve it, and that this could be profitable if their devices sold for comparable prices to next-gen TSMC tech, but this comes at considerable cost to power efficiency, places intense demands on the fab to customize process to fit design issues, and sucks money and time away from longer-term research to pivot away from DUV. They would spend the next ten years on the same trajectory as Intel 14nm, where they keep squeezing minor miracles out of the chip design in exchange for increasing power consumption and overspecialization. They'd keep getting tiny improvements, for years; meanwhile, Western EUV will also be improving, yields will also be slowly catching up, and costs will start falling off as the EUV nodes mature (look at TSMC wafer prices for 7nm at start vs today). The whole reason the major players are transitioning to EUV now is because they all recognize that, long-term, the future of DUV is unsustainable, and EUV is going to be less costly. Chinese fabs also know this. If China manages to stay competitive against EUV, they will be required to pay dearly for it. Lay-journo interpretations notwithstanding, this is, and has always been, the aim of sanctions - to penalize and slow access to cutting-edge technology.

I suspect that an invasion of Taiwan in which the outcome appears dire would result in the EUV machines being rendered into indistinguishable welded slag and scraps, at least in the molten tin UV source. Several hundred billion dollars and decades of research and development is tied up in the design and manufacturing of these machines. The critical trade secrets are going to be unrecognizably destroyed. If TSMC doesn't destroy their machinery, and ASML can't, the US will, covertly if possible, overtly if unavoidable. China will eventually get access to domestic EUV with enough will and investment; but without intact examples of working EUV machines to study, they're stuck on the long, slow grind of figuring it out themselves. How many years of progress will be made in western chip fabrication while the Chinese are busy re-solving EUV?

Inside the motte: maybe this is small of me, but it chagrines me every time I make it two paragraphs into a post and hit the pivot to antisemitic apologism or whatever, and only then realize I forgot to check the poster. I steadfastly don't care about any of that conversation, and it's the same conversation every time, so it's a few minutes wasted. I already know I'm going to ignore the post; blocking the single-issue poster would just cut to the chase, and would make me feel a little less dumb, on average once every week or two. That said, my blocklist is empty - it's such a minor nuisance, and feeling dumb once in a while builds character.

Outside of the motte (a selective, filtered, moderated community): to a first order approximation, the entire internet is spam, ads, and tribalism. See also: Most of What You Read on the Internet is Written by Insane People. My attention, patience, and cognitive filtering capacity is finite, and the internet is an infinite noise generator that evolves (read: gets louder and more annoying) by consuming my finite resources. If there are mods, they're usually crazier than the users.

I used to have a Tumblr. I blocked every significant nexus word related to politics, most of the fandoms topics, anyone who posted anything even slightly porny, and any specific person who found a novel way to annoy me. It was great!

I basically don't use Twitter, but I know a guy who blocksany person the instant they make a questionable post. His feed is a near-perfectly-curated niche technical community who go to the same conferences and share entertaining or interesting industry stories and papers. It seems pretty great!

Blocking is a crude tool, but it's one of the few scalable solutions to autonomously build a community or interest feed on top of a dumpster-fire social media platform.

Nice flair btw.

In industrial robotics, there's two ways you get consistency, reliability, efficiency, and speed:

  1. Limit the complexity of the required motions
  2. Control the environment to deliver the same inputs as often as possible

Manufacturing automation designs machines that break complex assembly problems into many separate sub-problems that can be solved by simple motions under a strict set of inputs. Much of the complexity of these machines is in developing schemes to guarantee the shape, weight, orientation, and velocity of pipelined precursors. Nearly all will "fail gracefully" under some unplanned set of inputs conditions at any stage in the pipeline - in other words, give up and complain to a human operator to fix the problem.

The value of AI in robotics is that it can help plan motion in uncontrolled environments. This motion could be simple or complex, but most examples you'll see are simple. For industrial robotics, this might look like machine vision and simple actuators to adjust orientation of inputs, or automated optical inspection to check for defects on outputs. But the whole value of automation is improvements over human benchmarks on the metrics listed above, and given the choice between designing a general purpose robot or a highly specialized machine, the specialist almost always ends up simpler, cheaper, and better at what its designers want it to do.

Self-driving cars are one of a small handful of applications where the mechanics are straightforward, but the environment is chaotic. The moving parts are all outrageously simple, even for racecars: the wheels tilt to steer, the wheels roll to accelerate, the brakes clamp to decelerate. The mechanisms that make each of these motions happen have a century of engineering behind them, of which many decades have been spent enhancing reliability and robustness, optimizing cost, etc. The only "hard" problem is safely navigating the uncontrolled environment - which makes it a slam-dunk next-step, since the unsolved problem is the only problem that needs focus.

The average blue collar laborer is combining dozens of separate actuators along many degrees of freedom to perform thousands of unique complex motions over the course of a workday. I have no doubt that advances in AI could plan this kind of motion, given a suitable chassis - but the size and form factor of manufacturable actuators with power comparable to their human analogues are physically infeasible to compress into the shape of a standard human body. Take a look at the trends in motor characteristics for the past few decades, particularly figure 8 (torque vs weight) - neodymium magnets and solid state electronics made brushless DC motors feasible, which greatly improved the power density and efficiency, but only modestly enhanced the torque to weight ratio. At the end of the day, physics and material science puts limits on what you can manufacture, and what you can accomplish in a given volume. And the kinds of machines we can improve - mostly motors - have to translate their motions along many axes, adding more volume, weight, and cost. Comparatively, human muscle is an incredibly space-efficient, flexible linear actuator, and while we can scale up hydraulics and solenoids to much greater (bidirectional!) forces, this comes with a proportional increase in mass and volume. This actually isn't so bad for large muscles like arms and legs, but for hands (i.e. the thing we need to hold all the tools) there just aren't many practical solutions for the forces required on all the required degrees of freedom.

In terms of what could suddenly change the equation, I suppose there are a few things to watch out for:

  • Room temperature superconductors could potentially increase the torque/weight ratio for motors by a whole fricking lot. This doesn't totally solve the problem, but it opens a lot of doors.
  • Cheap artificial muscles could help. There are a variety of designs available today, but most are science experiments (EAPs, thermal braids), pneumatics aren't exactly cheap, and rapid cycling of high pressures through tiny valves is challenging (not to mention loud). If anyone can make an electroactive polymer that's cheap, waterproof, and can torque, we might very suddenly be in business, but I'm absolutely not holding my breath on this one. The thermal braids tend to be really hard to control precisely because getting the fibers to a specific temperature is very challenging, and there's obviously environmental limitations, but otherwise they're potentially incredibly cheap and straightforward to make. Combining twisted fibers and electrical actuation could be promising, but I don't really see anyone doing that. There's a whole lot of material science between here and there.

My bet is on neither of these things happening any time soon. Basically every university in the world has an artificial hand or two under development, and they all suck. State of the art routinely costs six figures, weighs 5kg, and moves slow on 4x speed promo videos - it's been this way for decades and it isn't really getting better. Human hands enjoy a massive, durable nanomachinery advantage

Doesn't totally answer your question, but consider disabusing yourself of the notion that there's any chance of a robotics revolution in our lifetimes. Costs of robotics go up with degrees of freedom, power density, and sensory complexity. Human hands are the standard interface for all tools used by manual labor. Human hands have 27 degrees of freedom, can exert over 100x their weight, and can regularly sense micron (irregularly: submicron) texture. >75% of all non-industrial grippers in the literature can't operate tools with index finger trigger switches. The bare minimum requirement for replacing blue collar labor is making grippers with close to human hand functionality, mass producing them on a robot that can move and work anywhere a human can, and selling it for less than the cost of a fighter jet. You'll notice that I haven't even gotten to the rest of the robot yet.

If AI is any significant part of the next few decades, your new job will be physically laboring for it. There's rather a lot to do, and plenty of now-unemployed white collar workers to keep occupied...

There's a lot of documents, and it looks like the guy has been excerpting individual records as he encounters them to show them to various politicians. The whole tranche is currently up on his Twitter: https://twitter.com/SheriffLeaf/status/1769561564993192198

The documents show:

  • Source code modifications being made during deployment periods during an election - some firmware version needed to be changed, or some extra component needs to be attached at the last minute, or they suddenly need to disable parallelism on the SQL instance to keep concurrency bugs at bay, or whatever... They also talk about the software being able to receive OTA updates in places.
  • Serbian nationals accessing election infrastructure on election night (Dominion has had an office in Serbia for over a decade, where they employ many programmers; however, Serbia does not allow certain rigorous US background checks that should be standard for such employees. Presumably the access is related to bugfix work)
  • Emails that seem to indicate someone made an unauthorized access to Dominion backend systems several months before an election, from a suspicious location, on a company device or IP (extent or severity unknown, but may constitute a significant security breach?)
  • Some of the Serbia machines are behind on SQL server vulnerability patching by as much as two years (sort of a problem since their systems and testing would heavily utilize the affected SQL variants)
  • Numerous references to machines with modems installed, including some that failed acceptance testing for not having modems installed (again, plenty of good reasons for this in practice, but again, perjury)
  • The Serbians discuss sending election data over the internet as one of the use cases in a conversation about certificates
  • A long email chain showcasing the firmware programming process with a Taiwanese OEM (most computer manufacturing takes place in China or Taiwan, including firmware programming, so this isn't unusual, but it's laughably insecure)
  • Eric Coomer suggesting no VPN usage in Cook County systems for technical reasons (I like the proposed explainer where they talk about using unlicensed trial version of something and wiping the machine every 30 days to keep it installed) - seems like they eventually got a VPN up at the end
  • Eric Coomer opining on how Antrim county is full of angry conspiracy theorists, and also, gee whiz, how did that RTR misidentify the cartridges, I thought we fixed that
  • Testimony from some expert witness describing the aforementioned RTR misidentification, and how it could be utilized in conjunction with SQL database management software to trivially edit tallies without privilege escalation and avoid discovery at the end unless a total hand count was kept alongside the results (see https://twitter.com/SheriffLeaf/status/1769745766703374401)
  • Lots of "oh no, disaster, we suck" and "how can we paper over this mess to sell to customer" that I'm sure anyone in a customer-facing role at a tech company recognizes as business-as-usual
  • Numerous small firefighting cases where something didn't work right and Dominion has to cook up an explanation, occasionally beclowning themselves in front of the customer in the process, while everyone stands around waiting for the numbers to add up - a particularly noticeable instance revolved around some weird data duplication issue

It doesn't look like everyone (or anyone, really) at Dominion is twirling their moustaches and cackling as they disenfranchise the American people. It looks like they run like a standard tech company, which is to say all over the place, constantly fighting fires and doing the needful to get their sales. I'm sure 2020 was a complete nightmare scenario for these guys, where suddenly all their customers are radically transforming their deployments and doing novel, untested, gigantic-scale absentee and mail balloting.

But very clearly they're held together with duct tape and prayer in a lot of cases, which is about the opposite of what I'd like to see from critical election infrastructure. The glimpses of the architecture they have put together with all of these machines raises some significant doubts about the security of the enterprise, particularly if rank-and-file technicians can just go pop open the database manager and flip the counts around - surely this is the kind of thing that could be trivially accomplished if one could land a zero-day on any of the long-dated Windows 7 machines floating around in their customer base.

I'm told, with no particular means to corroborate this, that ES&S is about the same.