You’re right, shallowly — the quality of their implementation bears on these results.
One could read this paper as Salesforce publicly weighing their own reputation for wielding existing tools with competence against the challenges they met getting those tools to work. Seemingly they would not want to sully that reputation by publishing a half-baked experiment, easily refuted by a competitor to their shame? It’s not conclusive, but it is relevant evidence about the state of LLMs today.
I interpreted this to mean he holds clean water (e.g. from a bottle) in his mouth when entering the toilet, for the purpose of wetting toilet paper. A clever solution, IMO.
Certainly! The limit is your creativity. My first idea: generate very large scale 2D noise. Choose a threshold to divide it into regions. For each region, choose a direction of motion. Design a “mountain envelope” function that considers distance from the nearest border to create mountains/subversion zones based on the direction of each plate and the shape of the border.
In the second quote, the phrases "in their official capacity" and "as the position of the United States" are doing a lot of heavy lifting.
The EO is going out of its way to broadcast that its purpose is to establish a unitary policy position of the executive branch that stems from the President, rather than having "independent" agencies providing contrary position from within "in their official capacity" "as the position of the United States." The logical leap from there to "the President's (unrestricted) opinion is the law (without reference to Congress or the Courts)" is vast.
The EO does not bear on the balance of powers between branches of government, but on the ability for the executive branch to function as a single entity within that balance, rather than a multiplicity of quasi-"independent" agencies.
The disincentives that have always prevented the executive from blatantly violating the law are still in force and unchanged. They have been functional through 250 years of Presidents testing the limits of their authority.
> The disincentives that have always prevented the executive from blatantly violating the law are still in force and unchanged. They have been functional through 250 years of Presidents testing the limits of their authority.
Can you elaborate on what those disincentives are? I am thinking:
- Impeachment
- Charged with a crime, found guilty, sent to jail. It seems like this one is no longer possible due to Trump v. United States
- Killed by opponents
Without the criminal charges being on the table, those disincentives look a lot weaker to me.
Similar to Roe v Wade and Chevron, they'd need to overturn Trump v. United States and then charge him with crime (or contempt). So there is a ball in the judicial branch if they get pushed too far. Just extra steps.
The 4th one is us getting a retroactive watergate effect where republicans wake up and approval plummets. There's a non-zero chance Trump steps down if he pisses off everyone.
This only happens with some truly heinous actions that can't be spun. Like, bombing american soil or a ulta-blatant conspiracy with China/Russia. Or you know, him killing Medicare/social security (the most likely actions, given the budget proposal).
>> Similar to Roe v Wade and Chevron, they'd need to overturn Trump v. United States and then charge him with crime (or contempt). So there is a ball in the judicial branch if they get pushed too far. Just extra steps.
_Who_ would charge him with a crime? Prosecution is the responsibility of the Executive branch via the Department of Justice. The only option the court would have is Contempt, and I don't see that being particularly effective.
The only legal avenue this order leaves open is impeachment, and because that requires a 2/3rds majority in the Senate, there's all sorts of ways to prevent it. Even if the republican senators started to oppose him, the DoJ could be used to threaten and investigate senators who step out of line. Or a violent mob could be used to interrupt the impeachment vote.
>_Who_ would charge him with a crime? Prosecution is the responsibility of the Executive branch via the Department of Justice.
He only needs to piss off two SCOTUS's to start to have the entire book thrown at him. Unlike Congress, Judges tend to be somewhat resiliant to the political atmosphere. That's a lot more viable than congress. Less people to squabble with and people less concerned about losing their job. I woudn't even count out the entire DoJ on this either.They've been denying some of Trump's craziest EO's.
>Or a violent mob could be used to interrupt the impeachment vote.
Good. I'm tired of being blamed for "rebellion" everytime someone stands in front of a building with a sign. Let Jan 6th repeat while all eyes are on the Capitol looking at Trump.
I'm no legal scholar, but sure, I'll elaborate as a layman.
During the Presidency, impeachment and judicial review are the important checks. During the transition of power, state-controlled (as in, not federally controlled) elections and Congressional certification are the important checks.
While many people are disappointed with Congress' hesitation to impeach, impeachment as an institutional protection still works as a protection against blatant disregard of the law. Congress faces the threat of re-election practically constantly, and obvious disregard of the law without impeachment is not in Congress' best interest.
Judicial review is clearly working. I don't need to recap the large number of cases that have be brought against the Trump administration this term. The administration is abiding by stays/injunctions, and the Courts are issuing opinions independently. The Supreme Court, even Trump-appointed Justices, has ruled against Trump before and will likely do it again.
State-level control of Presidential elections and Congressional accountability to the public during certification protect against the spectre of a third Trump term or end of democracy. This scenario is extremely unpopular even among Trump supporters and would be a disaster for those in Congress.
The recent Supreme Court opinion about Presidential immunity in his official capacity explicitly defers and leaves to the lower Courts to determine the precise limits of that official capacity. It is beyond belief that the entire judicial branch would collectively enable dictatorship "bEcAuSe iMmUnItY".
That opinion specifically defers to lower courts to define the precise limits of the President's immunity. I refuse to believe that the collective effort of the judicial branch would construct those limits in a manner that enables dictatorship.
The lower courts can be appealed to the higher courts, so the opinion of the supreme court was effectively that the president is immune if they say so, on a case by case basis.
It was a clever but blatant way to give their side immunity without giving it to Biden
You're correct that the Supreme Court ultimately has final appellate jurisdiction on matters of immunity, but that's a long way from enabling dictatorship.
To get there, we would need to assume that the Supreme Court, including only 3 Trump-appointed Justices, is both unwaveringly partisan and unwaveringly supportive of dictatorship.
The Supreme Court is unwaveringly partisan and corrupt; its members have taken bribes and invented new rights for Trump to keep him out of jail. I wouldn't assume that they'd reject a dictatorship.
> TikTok is perhaps the most impressively addictive gambling app ever created.
> Geopolitics aside, I think everyone is kind of aware that gambling is a vice, and like it or not, this could just be the beginning of our society beginning to scrutinize these platforms.
The comparison here is a slot machine: you pay a a few to play, you pull the lever to play, you win a prize.
Here, the payment is your attention, you swipe to the next video to play the game, and the prize if you land on a good video is a small hit of dopamine.
Everyone's losing their collective mind about people watching videos on a platform not approved by our oligarchs, while there's an epidemic of people racking up gambling debt from the sudden prevalence of DraftKings and other mobile sports betting apps.
At least in circles I frequent, people are pretty upset with the state of sports betting too. Feels like lots of things are pretty crappy these days, simultaneously
The comments here are focused on how much energy it would take to turn this into fuel. The real story here is decentralized fertilizer production, buried at the end of the article:
> this innovation could fundamentally reshape fertilizer manufacturing by providing a more sustainable, cost-effective alternative to centralized production
The high energy cost of Haber-Bosch, plus the additional cost of transportation from manufacturer to farmer could potentially be eliminated by distributed, passive fertilizer generators scattered around in the fields.
I'm no expert, but assuming sufficient local production, low concentration could potentially be overcome by continuous fertilization with irrigation throughout the growing season.
Let's find out. Some quick fiddling with a molarity calculator and an almanac:
-- 100 uM ammonia -> 1.7 mg / L ammonia
-- 82% nitrogen -> 1.4 mg / L nitrogen
-- My lawn needs around 1 lb / 1000 sq ft, or around 5 g / m2
-- So my lawn needs about 3500 L / m2 of fertilized irrigation total for the season
-- Ballpark farming irrigation is around 0.2 inches per day, or around 5L/m2
I would need to water my lawn about 700 days in the year, or more realistically up my irrigation rate by about a factor of 4, AND source all of the water from the fertilizer box.
I'm a little skeptical that I can allocate space for enough production and still have a lawn left to fertilize. The tech probably isn't ready for the big time on an industrial farm yet, but for research demo, this seems like a promising direction! Much more than concentrating it for fuel.
So, farms are definitely setup already to accomplish this. Most farms have moved to central pivots for irrigation, and they already inject fertilizer into the pivot [1]. If fertilization could be generated onsite, then you could theoretically have everything plumbed together to "just work" without much intervention or shipping of chemicals.
Rain will wash nitrogen away (down to streams, rivers, and then the ocean creating lots of problems) so you want to apply nitrogen with an eye on when it will rain so your fertilizer stays on the field where you want it. Your link doesn't specify what fertilizer is being applied, I would guess nitrogen is not one.
Ammonia should be applied to the soil - in the air it is a hazard that can kill people and harm the plants (farmers wear lots of protective gear when working with ammonia, with more other things they don't bother).
As such I'm not convinced that is the right answer. You want a system that will apply nitrogen
Farmers use anhydrous ammonia that bounds with water in the soil and then bonds to the soil.
I don't know that farmers wear anything special when applying it, but there are safety procedures. I work with a farmer and he was telling me about one time he forgot to switch one of the valves off and when he disconnected a hose, the fumes knocked him out. Luckily it was just the fumes from the hose and not the whole tank or he likely would have died instead of just being knocked out.
Farmers already do keep an eye when it will rain before applying fertilisers. So, this is already part of their calculation. Although, yes , this means they will not apply it everyday. Depending on their location this means that a lot of weeks are out of the picture.
Cows and chickens cannot fix nitrogen from the air. They eat the nitrogen-fixing plants. So in a sense they don't "generate" fertiliser, they only concentrate it.
In Western Europe at least, cows are kept in pastures, which are permanently dedicated to that use and not used as fields to grow crops. All in all, cows are probably only low carbon in premodern rural contexts in West Africa and Asia for instance.
Of course you can't have cows wandering through your corn or soybeans, they'll eat and/or crush it. But if you had fields that you could rotate between pasture and planted that could work.
ANFO is explosives made with ammonium nitrate(Ammonium Nitrate Fuel Oil), however ammonium nitrate is by itself rather energetic and will explode when store improperly. The most recent memorable incident would be 2020 Beirut: https://en.wikipedia.org/wiki/2020_Beirut_explosion
Imagine one of these units left somewhere, slowly filling a tank that has not been sealed, water evaporating back out leaving a nice ammonium nitrate powder behind....
NGL, it would be an easy sell. You are just a hop/skip and a quack away from turning that decentralized fertilizer into a decentralized bomb making system.
A hop, skip, quack, jump, and fairly obvious high-energy distillation process away. The national security angle probably isn't a concern here for the same reason that this process doesn't produce good fuel.
Ammonium nitrate is made from ammonia and nitric acid (which is also made from ammonia). Therefore, ammonia is the only necessary direct precursor to ammonium nitrate, which is probably the most relevant oxidizer in improvised explosives today.
Not saying that it should be regulated on the basis of national security, but it’s not like there isn’t a potential security concern.
It's not out of business. It merged with Bayer. It's a change in ownership, and to some degree a change in upper management, but large swathes of the company are unchanged.
Its assets were sold to Bayer and BASF and some former Monsanto workers may have begun working at those other businesses, that is true. That kind of scenario is true of all businesses that close down, though, at least unless they truly have no remaining assets to sell or workers wanting new jobs, both of which are unlikely for anything beyond the simplest of sole proprietorships. By your logic, there is almost no business in history that has ever gone out of business.
I mean, you can vertically scale these, right? It isn't solar powered.
Also while I understand the local delivery aspect, why waste arable soil on this, use desert. Hm, do you need moist air? They were testing it near water. Maybe on the Pacific side of the Rockies? That air is moist enough to drop feet of snow at once.
It's a recent use. I'm still not convinced it's a good use case. I think it's mostly greenwashig (bluewashing?) to avoid the explicit release of CO2, but probably biodiesel is a more ecological friendly alternative.
> Many good ideas have been destroyed by too-rigid applications
This is an interesting perspective that I disagree with. You seem to be saying that a general misunderstanding or misapplication of a good idea degrades the idea to a point that "destroys" it.
On the contrary, I believe there is experience to be gained in practice: either the initial idea wasn't good after all, in which case we're destroying an illusion; or the poor examples serve to refine the idea by clarifying some ambiguity that, interpreted wrongly, leads to failure.
Perhaps your argument is that many people may become familiar with the bad implementations and the idea's popularity will decline, depressing demand for refined implementations. This is likely true, but reflects the tragedy of the anticommons, not a degradation in the idea itself.
Well, what I mean, is that folks invest in the dogma, to the point where they refuse to accept any changes to fit realities, and often believe that they can apply the dogma, in areas, or in a manner, where it is not appropriate.
There's an old Swiss Army saying "When the map and the terrain disagree, believe the terrain.".
People who invest in dogma, refuse to look at the terrain. The map is The Only Source of Truth.
I believe that most dogma comes from something that works in one or more contexts, and may actually be highly effective, in other contexts, as long as it is adjusted for context. That last part, is what kills it. People refuse to change, and the dogma gets a bad name as a "failure."
This is an interesting take, the article touches on it too.
> “Physicists are much less concerned than mathematicians about rigorous proofs,” says Timothy Gowers, a mathematician at the Collège de France and a Fields Medal winner. Sometimes, he says, that “allows physicists to explore mathematical terrain more quickly than mathematicians.”
GP is right that the currently observed physical laws go far beyond our ability to observe them in reality because of the cost of observations. International effort over decades is required to create facilities capable of making helpful new observations: think of LHC, LIGO, James Webb, etc.
On the other hand, once the facilities are built and ground-breaking observations appear, we suddenly have a debt of theoretical and simulated exploration to understand all their implications. The low cost of computation greatly extends the value we can take from every truly new observation of reality.
In order to observe something new, we must be able differentiate it from something already understood. It seems like the physics and math communities are currently in a season of increasing our understanding of the existing models well enough to motivate trying to break them.
One could read this paper as Salesforce publicly weighing their own reputation for wielding existing tools with competence against the challenges they met getting those tools to work. Seemingly they would not want to sully that reputation by publishing a half-baked experiment, easily refuted by a competitor to their shame? It’s not conclusive, but it is relevant evidence about the state of LLMs today.