I think this is not the correct way of thinking of Moore's law. Like mentioned elsewhere, Moore's law is not so much a law as a self-fulfilling prediction. Semiconductor companies follow Moore's law (or recently an approximation) to stay competitive because their competitors do it and hence Moore's law continues. The cost of not marching along to Moore's law can mean you will get left in the dust (see Intel).
I think a better question to ask is whether the underlying economic factors behind continuous process tech improvements are healthy. Is there enough value-add to the final user by continuous process tech improvements? Are the costs for that improved process tech scaling with the value-add? And is the competitive landscape healthy? While that holds true, companies will keep looking for process tech improvements to give them a competitive edge.
In the 80-90s this was very much true but in recent decades it was to a lesser extent hence why we see consolidation/reduction in the number of foundries, foundry services to amortize the cost of older node-tech, and R&D going to companies/partnerships that can capture the most end user value-add (Apple/TSMC).
Looking forward I think the economics are very healthy with a design-house/foundry service model that we have right now so I would guess that Moore's law (or some approximation) will continue for the next decade. There are a lot of process tech innovation that can lead to better performance that are not necessarily scaling related. In fact, scaling transistors stopped being very useful a while back afaik due to the breakdown of Dennard's scaling.
>"I think a better question to ask is whether the underlying economic factors behind continuous process tech improvements are healthy. Is there enough value-add to the final user by continuous process tech improvements?"
I agree with this and I suppose I just sort of reached for Moore's Law out of habit and maybe a bit of laziness. Thanks for articulating the question more appropriately.
>"Looking forward I think the economics are very healthy with a design-house/foundry service model that we have right now so I would guess that Moore's law (or some approximation) will continue for the next decade."
I am not sure toughness has anything to do with ferromagnetic properties. But it seems like a bad idea anyways considering that very few materials are naturally ferromagnetic.
Very high field magnets are critical for MRIs, tokamak fusion reactors, mass spectrometers, and probably many other uses if we consider pure science as “useful”.
Actually particle accelerators somewhat funded the early investment in high field superconducting magnets which might now appear to yield improvements in fusion energy Q factors due to the strong dependence on field intensity [1]
Of course it has to be useful. Based on that we build our world (through engineering) and if we work for something that cannot be used then it does not make sense. I do not see any use for quantic foam or quasars any time soon.
As for magnets: this certainly helps, but has CERN done any breakthroughs that were used afterwards? They had at some point the world record for the density of a magnetic field. There is no practical use for that neither in MRIs (that require a lower field and were used for many years already), nor tokamaks (where the homogeneity of the filed is paramount on larger scales compared to the ones in CERN.
I am not saying that the engineering work that is done in CERN is useless - it is just that the money poured there goes primarily in some "whose is bigger" contest that has exactly zero chances to be used.
> the only things you can know are the things how they actually are
Yes, but investing MM€ to know something that has no use does not make sense, if there is much more important knowledge competing for the fund. I guess that knowing how to cure MS is more important that discovering a four quarks particle, right? In an ideal world we could do everything but we have to choose wisely because the amount of funds is limited.
> A PhD from CERN should not confuse science with engineering.
Agreed! The list of highest growth and highest decline cities by page view is almost entirely small cities. This is a pointless stat because of course the small cities are going to see the largest change from even the smallest of trends.
I’m skeptical of the rest of the analysis if simple things like that are not considered.
Any location estimator worth its salt is already doing this through the use of a multi-rate EKF. On top of that, it takes into account the vehicle dynamics (i.e. cars can't move perfectly horizontally) to improve estimates.
The "novelty" here is the ML approach although I am not sure if that is particularly novel as well.
Exactly. I find it exceedingly odd that well-understood techniques like double integration and Kalman Filtering are being replaced by ML black boxes in the hope that ... what? They're going to rediscover what we already know? But in a format where we have no hope of figuring out what they discovered, so we can never bound their domains of acceptable performance?
Sometimes there are other regularities in the data that an ML algorithm can exploit; for example, if it can detect feet strike events in the IMU stream, it can compute an estimate of how fast someone is walking that is independent of integrating accelerations.
The benefit of ML over long-standing algos like kalman filtering for integrated navigation systems is they can can give you the occasional surprise when an unexplored edge case happens. People like surprises.
I've used ML to supplement straightforward heuristics before.
For instance, I built a fuzzy matcher for business addresses which was based on hand-rolled heuristics but used logistic regression to train a probability estimator that the match was correct and then hand-evaluated a few thousand matches so that the matcher knew the quality of results it was returning.
It’s also incorrect to assume the expected value of the equity is zero. It’s much more useful to model it as E(x)=x*p(x), especially if you have some useful information on estimating p(x) that a person working at a startup might.
Every startup thinks their company is going to be successful. No matter what the person thinks who is working their, they don’t know when the investors will cut their losses or when the market isn’t hungry for money losing startups and the VCs have to put off going public…like now.
An hour of PBS Newshour a day. Very much fact based reporting with maybe a slight left leaning bias. Digest it and move on and live your life.
Besides that, I just turn off the news spigot. I'm slowly coming to the realization that watching news was kind of becoming my version of watching reality TV shows. There is of course value in being informed but outside of that, a single person can't meaningfully take action on all the things being reported on. So I see less and less reason to keep up with the news besides just being generally informed.
I think a better way to consume news might be to get a few "general" news tidbits through your regular news outlets. And then get more focused news on topics you personally care about and are willing to take action on through non-traditional news outlets.
As someone with a conservative leaning bias, I would consider PBS to have a very strong left (whatever that means) leaning bias. I guess it depends on your start point.
As someone who puts priority on distinguishing rhetorics vs facts, PBS is the furthest of the mainstream news outlets toward the "facts" end of the spectrum by a very large margin.
Is there anything concrete you can say about that bias and PBS News Hour?
Or is it more just general "they're publicly funded and the left supports that" sort of bias? Are they disproportionately covering or ignoring issues due to that bias?
I have one example. They were doing an article about why it is bad to legalize pig hunting. They quoted an "expert" who said "I've seen deer hunters, and they LOVE hunting deer... so I would expect anyone with a gun to release pigs in other people's land so they can hunt them"
I happen to know several hunters, and that kind of accusation was pretty offensive and far from what actually happens in real life.
This is the type of thing to overlook as being bias when you don't grow up in that culture that would be offended at the accusation. I watch the PBSNewsHour, but I absolutely recognize they have a bias on certain topics, and know I am unable to see the bias because I am too steeped in it on other topics. That is true no matter how "fact based" any reporting is. It isn't terribly hard to stick to "just the facts" and still leave readers/watchers with a very warped picture by putting greater emphasis on some facts while downplaying or outright ignoring others.
Yeah maybe. Are we talking about PBS overall or the Newshour segment?
In PBS Newshour the upfront report by Judy is very much fact based. I agree though that the various mini-documentary parts of the segment are less direct-fact based reporting and bias starts to creep in.
Anyways, I'm not too interested in starting an internet politics fight. The main takeaway from my post is that we should reevaluate what we are getting out of following the news obsessively.
It's a completely BS statement, so I don't know what you're trying to say. It's not even possible for reality to have a bias in the first place. Reality just... is. It's the people who try to tell you what reality is that have biases.
It's glib, but the meaning is that rational observation tends to strongly disfavor conservative views. Things like climate denial and supply-side economics that are tentpoles of conservatism consistently fail to hold up to any level of scrutiny. Modern American republicans are currently hanging their hats on completely farcical assertions that the 2020 election was stolen and Democrats are all pedophiles. When an outlet like CNN reports the truth, they are going to appear liberal in comparison.
That's not what it means. It means that rational observation free of bias aligns more closely with liberal politics.
Per the examples given: Scientific observation says climate change is real, man-made and an urgent threat. Liberals agree, conservatives don't. Maybe not to a person and maybe not with 100% fidelity to the science but the trend is unmistakable.
Same for many other conservative fallacies like "tax cuts pay for themselves". We have data and know it isn't true. And current hot topics like election fraud. All the evidence says it's extremely rare. Liberals agree with the evidence, conservatives don't.
That's not the entirety of the political divide and there's plenty of subjective and philosophical arguments irrespective of evidence but when it comes to justifying policy, one side relies on evidence one doesn't.
No, both sides have different realities they don't like-
Reality Progressives don't like:
- Gender correlates with biology
- Giving chronically homeless people housing won't fix their homelessness
- Not charging serious crime won't lead to safer neighborhoods
- You can't fix societal disparities by changing who is at the top and bottom of a pyramid of privilege
- Demonizing the wealthy won't lead to any productive outcomes
- The amount of GDP consumed by government already should be manifestly sufficient to fix the things government is capable of addressing efficiently
- Universal Basic Income is a pipe dream with no mathematic or social basis in reality
- Getting rid of religiosity without replacing it with another moral foundation will lead to progressively worse societal outcomes (and progressive thought is in no way an adequate replacement)
- Shouting down unpopular opinion doesn't make those opinions go away
Reality conservatives don't like:
- Locking prisoners away for long periods leads to greater gangsterism and lifelong criminality
- Racial disparities, whatever their source, need to be addressed systemically for society to prosper
- Making huge changes to our atmospheric mix is 100% likely to lead to undesirable outcomes- the globe, and humans as part of it, will not prosper with large changes leading to unknowable outcomes in our biosphere
- Education is essential to social mobility
- Social safety nets are essential to social mobility
- Free markets are rarely free of defect and tend towards capture, either regulatory or monopolistic, and need frequent intervention to function efficiently
- Even if you've felt deceived by the media, making it a point of pride to doubt anything/everything you hear opens you up to those who will manipulate that doubt for their profit and power
Either way, the smugness both sides have in being sure they are the 'righteous/scientific' side only opens them up to lack of self-reflection on whether their side is correct on a particular issue. If you're sure that your 'sides' agenda is correct top to bottom, you've assuredly sold yourself a bill of goods.
Nothing you listed under progressive misapprehensions are liberal dogma, nor are they categorically disproven. No sane person would argue that biology has no bearing on gender identity. The liberal idea is that people should be able to live their lives as they choose. Obviously someone with XY chromosomes can never get pregnant even if they outwardly change their gender. Giving the chronically homeless housing has had some positive results and nothing else has, so it's something some democrats have been willing to try. It's experimental, but definitely hasn't failed and we would absolutely abandon it if it ever does. Absolutely no one advocates not charging serious crimes, that's propaganda. UBI is also nowhere near liberal dogma, but has some credible theory behind it worth exploring. We already do EITC and it's wildly successful. The democratic candidates for president in 2020 who supported UBI got zero delegates. Government has enough money to solve everything it should solve? That's just unquantifiable gibberish.
Regardless, my point is still that the preponderance of evidence aligns with a preponderance of liberal policy. Climate denial, covid denial and election fraud conspiracies on their own are enough to condemn modern American conservatism to the garbage heap of corrupt populism even if liberals were buying every drug-addled hobo a luxury condo. I'd still rather live in that world.
AND it seems to be based on a simulation which can be fudged in all sorts of ways. Nvidia’s track record with cpus is spotty at best. Not sure how much weight to put on these claims, at least until there is a shipping product.
> Nvidia based this claim on a pre-silicon simulation that predicts the Grace CPU at a score of 740+ (370 per chip).
That is possible, but I think strains credibility. The GOFAST video is egregious, I find it hard to believe nobody in the US Navy was able to calculate the actual [mundane, very... balloon-like] speed of that object using basic trigonometry. I think it's more likely the Navy knew there was nothing actually exotic in that video.
The Defense Department described the subject of GOFAST, Gimble and FLIR videos as "UAV, Balloons, and other UAS". 'Balloons'; they know what it is. They're not even lying, they're being coy or misleading.
i think they are, there is a video of a "fighter pilot" who does not understand basics of a cameras... he thinks you can't focus on objects at different distances at the same time.... how did he ever get to fly a jet? or parallax i mean he should know how that works
I think a better question to ask is whether the underlying economic factors behind continuous process tech improvements are healthy. Is there enough value-add to the final user by continuous process tech improvements? Are the costs for that improved process tech scaling with the value-add? And is the competitive landscape healthy? While that holds true, companies will keep looking for process tech improvements to give them a competitive edge.
In the 80-90s this was very much true but in recent decades it was to a lesser extent hence why we see consolidation/reduction in the number of foundries, foundry services to amortize the cost of older node-tech, and R&D going to companies/partnerships that can capture the most end user value-add (Apple/TSMC).
Looking forward I think the economics are very healthy with a design-house/foundry service model that we have right now so I would guess that Moore's law (or some approximation) will continue for the next decade. There are a lot of process tech innovation that can lead to better performance that are not necessarily scaling related. In fact, scaling transistors stopped being very useful a while back afaik due to the breakdown of Dennard's scaling.