Sounds like Newton. Long tradition, I guess

Your use of the word "most" did not go unnoticed, and I am fully aware that the plural of "anecdote" is not "data." With that said, I have met plenty of people just like me in my career. I do not believe that "most" people running Windows shops" are "stupid and incapable." Some certainly are, but I've met plenty of dumb *IX people, too, especially back in the 90s.

I've met my share of assholes, too. Just saying.

"Okay, nuclear power. Yeah. Yeah, they did give us that. That's true. "

"And industrial-scale superconducting magnets!"

"Don't forget GPS satellite relativistic corrections!"

"And radioisotope therapy!"

"Electron beam therapy?"

"World Wide Web, wasn't that created by particle physicists?"

"Yeah, they did do that."

"Capacitative touchscreens too, you know."

"All right, all right, FAIR ENOUGH! But APART from computers, telecommunications, nuclear power, PET scans, MRIs, satellite communications, superconductors, radioisotope therapy, electron beam therapy, heat-shrink tubing, cargo and baggage screening, smoke detectors, muon tomography, capacitative touchscreens, and the World Wide Web, WHAT HAS PARTICLE PHYSICS EVER DONE FOR US?!"

"Synchrotron light sources for drug discovery?"

"Oh, drug discovery?! GROW UP!"

Well, part of the answer is clearly "because the official stance of the US govt. is that AI should not be regulated". (Yeah, it's more nuanced than that, but that's the short form.)

Come the fuck on.

In my career, I have worked (professionally) with Solaris, AIX, HP-UX, Linux, OS/2, SCO ODT/OSR, a bunch of shit I can't remember, and every Microsoft server platform since NT4. My current environment is 98% Microsoft. That doesn't make me stupid and incapable, that's just where my career took me.

Is... English not your first language? I ask because you appear to be way overthinking this.

The metaphor was perfectly clear to me when I read it.

Yeah, what has a better understanding of particle physics ever done for US, huh??? ... ...

_ ...........

if this were true it would not be possible for the system to read the watermark.

You confuse "invisible to humans" with "invisible to statistical analysis". Humans can't, for example, see the difference between metameric colours. Statistical analysis can. Humans can't see minor drift at scales where Brownian motion dominates. Statistical analysis can. Humans can't see the difference between background disease rates and new rising threats. Statistical analysis can. On and on. Statistical analysis is a far more powerful tool than human perceptual abilities.

this is also a lie. Identical thoughts must be expressed identically, otherwise they are, in fact, *different* thoughts.

I literally gave an example above. If you saw "picked a mango", "selected and picked a mango", "from within the tree, picked a mango", you would never think there was anything going on in the choice, and it isn't going to change the story one iota.

Tesco's - one of the UK's largest supermarkets - came out a few months ago and said "We've migrated off VMWare because of the licensing". They did it in months. On the major backends running every store in the country.

They had a massive dispute with VMWare and are suing them (they say they bought a perpetual licence, VMWare then cut them off), but while they did that... they just migrated their ENTIRE infrastructure.

Sure, it's a big thing to do. But it's not impossible, doesn't take years and - like Tesco - you don't need to wait to do it. If it's THAT business-critical? All the more reason to have an alternative ready to move to at a moment's notice.

Companies CHOOSE to get tied in, they CHOOSE to only ever run one system, they CHOOSE to make the migration painful, they CHOOSE to hold back on it in the hopes they can negotiate to carry on just as they are. And those are all DUMB from a business-continuity point of view and no substantial organisation should be allowing that to happen.

Pretending that huge organisations paying Broadcom millions are somehow powerless in this is actual complicity in the worthless side of IT where everything is just absent of any redundancy, resiliency, independence, failover, etc. etc. etc.

About 25 years ago, I was at a bbq where I spent the evening chatting with this guy named Lars Knoll. Even though I was busting my butt with my team to develop the Opera for Linux web browser, this guy had been hired by TrollTech after he had developed something called Konquerer which was becoming the default HTML rendering engine in the KDE project. He wrote the whole thing on top of the KDE libraries which were painfully slow. He rendered text and for controls using the KDE objects which were rarely more than Qt objects with a little KDE specific stuff on top. I remember thinking "there's no point thinking much about this because there's no chance it could compete with a real browser like Opera".

Shortly after, I was in Cupertino sitting across from a strange product manager type. His name was Tim Cooke. We were discussing developing the default web engine for this new Mac OS X project which was coming along. We would have to develop for Carbon because Apple hadn't implemented Objective C++ yet and we were a C++ shop and we couldn't use half the developers in the company to make a C abstraction layer for Cocoa. We came to an agreement and Apple sent us a palette full of the latest G4 macs and screens. I remember fighting over the 21" screens like children. We had an idiot project manager who insisted that we implement the browser the way Adobe was demanding. It didn't matter that we were pretty much the "experts" on making cross platform embedded browser engines. He was like "Adobe said it... it must be done this way... they are gods".

Well long story short, while we were implementing a crap Carbon based browser engine using crap tools from Metrowerks instead of the good Apple tools (which weren't that good yet). Apple put together a team which abstracted the KDE libraries and built bindings to C and then Objective-C/Cocoa. And they abstracted out everything that was platform specific otherwise from KHTML (Konquerer's underlying html widget) and wrapped it in a Cocoa wrapper and a Carbon wrapper and shipped it as Webkit and Safari. And man did it suck.

Who would have ever imagined that Lars Knoll's half assed KDE web browser would have eventually become the defacto Web standard and the default web browsing engine in every operating system... including Windows... except the strange 5 user corner case operating systems.

Don't bash a project like this. It has value. I immediately considered that I have had many cases in recent times where it could have been of value to me. That said, I wouldn't run a C based web browser engine ever. It's not that you can't write a good web browser in C. It's just that it's the wrong tool for the job. At 160,000 lines, it's a small job to port to something else. When JavaScript evolves into a JIT and the JS runtime is thunking the heck out of the DOM, then it's pretty much no longer an option. Oh, and someone will think it's a good idea to implement proper form controls. That's an insanely massive project, especially when adding WebGL as well. If this browser core catches on, which unfortunately due to the license, I don't expect it will, it will likely be 5 million lines of code almost overnight.

Yeh, no.

Three major events started "trade wars" and the government propaganda anti-China rhetoric.

1) 5G happened and Huawai was the only company with a full 5G implementation including millimeter wave and SA mode on release. Ericsson was 2-3 years behind. Nokia was 6. Huawei could deliver on day 1 and they cost much less because neither Sweden or Finland are ... well, they are terrible at competing. They simply expected China wouldn't be in the competition because... Why would they be. So, they were confused because they never had to actually compete against a company who actually did a good job.

Keep in mind, by now, nearly $2 trillion has been spent building 5G networks. Money to Sweden and Nokia recirculates. To China, it's horded and weakens the US.

2) Apple evaluated CXMT as RAM suppliers and YMTC and certified them as good enough for Apple to trust in their products and with their reputation. The US slammed the door shut on that. They are still making up excuses for why as there were no legitimate ones except that if those two companies get a foothold, the entire world semiconductor commodity trading market would collapse. As a result, non-Chinese companies can't legally license their tech to CXMT or YMTC so those companies operate patent free. But other companies have to pay license for their tech. This allows CXMT and YMTC to sell RAM and Flash for half the price. If the US lifts sanctions, in 3 years or less, Samsung, SK Hynix, and Micron could be considering bankruptcy of their RAM divisions and CXMT could aggressively dump RAM to collapse the value of the patents.

3) China focused on brands. Before Chinese brands happened, buying expensive Chinese products was a high risk, Huawei, Xiaomi and others started legitimizing Chinese premium brands. This made it so buying an $80k Chinese car was ok because it wasn't cheap Chinese no name crap.

The money floods to China and they invest in One China by spending obsessively on education and R&D. They raised 800 million people out of poverty so far. They know every yuan spent in China is worth it. They invest massively in Silk Road because they don't think 4 years at a time. Building infrastructure across Asia, South America and Africa will pay them back massively.

The one thing which will cause China's greatest growth will be when they start opening Chinese universities to the masses. It will be way bigger than any other Silk Road initiative. And it will devastate the US economy. The US cannot afford to lose access to the world's top talent and can't afford to lose control of who gets an education and who doesn't.

Did you know that China's defense against American anti-Chinese propaganda is simple pro-China marketing? Which do you think is working better long term globally?

There are people who see what's happening and prepares for the changes. There are fools who make idiotic comments like yours.

Exactly.

That line is clearly considering the 3.5mm jack as an "input" rather than an output - "...audio coming through the wired connection"

There were various Amazon devices with 3.55mm jacks - some were output only, some were input only, others could have their port either be an input or an output.

Source: https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.amazonforum.com%2Fs%2F...

How silly - you can carry hundreds, if not thousands of electronic books in a device that weighs a pound or two, how many books can you carry with you onto a plane? Into a coffee shop, etc?

("It actually makes RAM needs worse" -> "MoEs actually make RAM needs worse")

It actually makes RAM needs worse than a dense model, for a given quality.

That said, inference techniques like FreeToken are improving MoE swapping performance, so it's not as bad as it once was. Honestly, I would not be at all surprised if we start doing training in a swap-aware manner, where expert cache misses count as loss and models can queue experts to start loading before they're needed. Could even train in two tiers of loading - RAM and disk. Wouldn't that be great? Could run multi-terabyte models on consumer hardware. Training would tend to tend to concentrate key logic and reasoning on small number of very active experts, mixed reasoning/knowledge on less common experts that are usually kept in RAM, and rarer knowledge on experts that usually remain on-disk until needed.