So we know when AI trains on data trained by AI, the LLMs become more and more unstable. (Source)

Meaning the problem is not just "Social Media will suck more." It also means that a large treasure trove of data used by AI companies to train their bots will become increasingly toxic. And this will hurt the value proposition of companies like Reddit (which depend on selling their data for training AI), as well as companies like OpenAI, who needs more and more data to train on.

Investing in what, exactly?

I mean, when you invest in a company, you're investing in the people, the processes, the products of that company, on the idea that the hard work of those folks will lead to gains in your portfolio. But if there area no jobs because they've all been replaced with AI, what is left? Some sort of weird gambling casino where we're betting on the next genius idea that AI then implements for us?

What sort of dystopian bullshit future is this?

Of course it's from a company which manages people's investments and makes money off their trades, so I can see how when all you are is a hammer manufacturer everything is a nail that needs to be pounded.

Decades ago, the move to 'open offices' was driven by bad research. That is, the research on productivity was done with college students; they found that college students in a collaborative environment do better than college students trying to study in isolation. Which--well, that makes sense, given that college students are still learning, and it helps to have some collaboration while in a learning environment.

But that research was used to justify the whole 'open office' movement--forgetting that people like software developers are not college students, and need a way to drown out the 'forced collaboration' in order to find a modicum of peace so they could focus.

Of course, open offices aligned with managers who wanted to be able to see all the veal in the cattle pens workers working for them, and it aligned with the penny pinchers who didn't want to build enclosed offices.

And it was only decades later that we "learned" the painfully obvious: that open office floor plans are a failure.

And now we're doing the same damned thing with "hybrid work" and forcing people back to the office.

Both civic leaders who want to bring workers back into the downtown corridor so they have the captive audiences for commerce in a downtown corridor, commercial real estate owners who want full buildings so they can guarantee returns on their investments, and managers who want to see full veal pens their workers so they can 'manage' them, have all aligned with this idea that "returning to the office" is better, somehow.

And now comes the research--undoubtedly being done on college students, who in fact do benefit from collaboration. And not on workers who benefit from quiet space so they can concentrate on their work.

Worse, because of the absolute mess done by the pandemic shutdown requirements--and how people moved across the country (because they could), the push to get people back into the office is often accompanied by confusion and worse: a lack of desks for workers to work at. But we're ploughing ahead anyways, regardless of the loss of productivity or the loss of good workers--and I'm sure research will be "discovered" which support all of this.

And a decade or two from now, after the wreckage is done, someone will point out that maybe all of this wasn't a good idea: that the increased carbon footprint of daily commuters to fulfill some sort of financial and political obligation to large commercial real estate owners, as well as satisfying the need to fill veal pens, may not have been the wonderful idea prior "research" suggested.

The problem as I see it is that if we take it read that all of this is 100% absolutely true and we have three years to prevent an existential world-wide crisis from destroying the environment and rendering the Earth as inhospitable, then all eyes should be on China--which thus far has been exempted from various intergovernmental requirements to reduce CO2 emissions. And part of the problem is that while China itself claims it has capped carbon emissions--a lot of the environmental and governmental statistics coming out of China is as honest as a three-dollar bill. Meaning unless they are willing to actually audit China's numbers--which account for 31% of emissions, give or take a hell of a lot of uncertainty--we can convert the US and Europe to eating bugs (to save on cow farts) and it won't matter one whit.

But despite China's "Double-Carbon" policy announced in 2020, I'm personally not convinced China is doing anything outside of trying to position itself as a "responsible world leader"--that is, it seems China is more interested in displacing the United States and eradicating the "western centric" world than it is actually doing anything that is transparent, verifiable and actionable.

China can't help it, and for the same reasons why GDP numbers from China are untrustworthy: the CCP must maintain appearances to maintain legitimacy--which puts a lot of incentive on lower party officials to lie, cheat and steal.

And the real question is can the world, when facing an existential crisis, afford to put up with China's lack of transparency and unwillingness to allow outside observers actually audit all aspect of China's society for environmental compliance?

Before you think "that's stupid", remember that a lot of the environmental excesses of the Soviet Union were not uncovered until after the Soviet Union collapsed. And despite better intelligence gathering and satellite coverage--it's still possible to hide a hell of a lot behind government secrecy.

Like we did in Area 51.

I learned to touch type in high school--at a time where I had to fight the administration to allow me to take a class "for girls" to learn a "secretarial" skill. (Early 80's, but the teachers there were still stuck in the early 1960's, apparently.)

I find it incredibly useful to be able to express my thoughts and ideas without having to think about the keys or to look at where the keys are. Just sit my hands until my index fingers feel the little bumps, and away I go. And as a software developer it helps to be able to express complex code without thinking about where the keys are; in fact, I got rid of a keyboard because while I loved how it felt and how it looked, the back-tick button was moved down and to the left, and the escape key was placed to the left of the 1. And I found typing code escapes in Markdown and bitwise negation in C a pain in the ass; every time I'd think 'code' my finger would press the escape key where the backtick should have been.

Sorry, but having known some VCs in my own life, this feels like a significant disaster in the making as VCs who bought into the hype of AI dump a bunch of money on existing companies, fail to upgrade them, then dismantle those companies: old school 1980's corporate raider asset-stripping.

There are places where AI can definitely help--but I sincerely don't trust VCs to know where that is. I mean the whole 'vibe coding' thing came from the VC world--and in a sense, VCs don't give a shit if it works or not, so long as it maintains a high volume of churn.

If it were that simple and that impactful, economists would surely have adjusted the metric.

As to why we still use them: because they're easy to capture, they're easy to understand, they've been standardized over the decades, and they're politically useful. And no economist in their right mind uses the number in isolation; it's usually used along with a broader notion of the balance of payments. And when comparing trade deficits year over year across different economic sectors, it does give insight into how the economy is evolving.

But for the purposes the Trump Administration (and prior administrations; notice trade deficits has been on the political radar for decades now), the way the metric is being used is--to put it politely--faulty.

No... none of the money Apple pays to China for the manufacturing of the device goes into Apple's pockets.

What I mean by this is that suppose you order an iPhone from Apple for $1,000. It gets drop shipped from China to your home.

The imported item, the $1,000 iPhone, is counted in our trade deficit as the full $1,000 going to China. That is, we calculate the trade deficit by looking at the declared value of the imported item as it crosses the border from China to the United States, and in the case of your iPhone, we presume we just lost $1,000 to China, never to be seen again.

But here's the thing: Foxconn, the company who assembled the iPhone and who drop shipped it to you, only gets $50 to assemble and ship the phone. Yes, that's $50 that Apple doesn't get--Apple pays $50 to Foxconn. And note that there are other costs that go to other companies: Apple pays money to TSMC to manufacture the processor, they pay to Samsung for the display, they pay to other companies around the world for the other components--including money to other Chinese manufacturers.

But at the end of the day, even though we count the full $1,000 value of the iPhone as a trade imbalance with China, the reality is companies all around the world got a small piece of that $1,000 total cost.

That is, a Taiwan company got money for the SoC, a Japanese company got some money for the camera module, a European company got money for the gyroscope technology, a South Korea company got money for the display, etc., etc., etc.

And Apple makes almost $500 in profit.

In other words, and this is my point: while we credit the full $1,000 as a trade imbalance to China--half that money actually winds up in Cupertino.

But wait! Our trade imbalance numbers are even worse than that! Many of the items that are in your $1,000 iPhone are technologies manufactured by, or designed by, other American companies who get a substantial profit from those compnents. Qualcomm (in California) licenses the modem, Corning (out of New York) provides the glass, Micron (out of Idaho) provides the DRAM/NAND memory core, etc., etc.

So while our trade imbalance credits the full value of the $1,000 phone that was just drop shipped to you from China after you ordered it from Apple, it's quite likely somewhere around 35% of the actual build cost (the $500 used to make the phone) flows back to America. Meaning that while our statistics suggest we just had a $1,000 trade imbalance with China, the reality is perhaps $700 of that stays in or flows back to America.

This means the trade imbalance statistics are **WOEFULLY** inaccurate in representing the real world.

The problem I see is that the way we measure trade deficits don't account for the flow of wealth due to the value of American intellectual property.

Consider, for example, that in the trade deficit metrics, China gets full credit for the $500-ish import cost of an Apple iPhone. That, despite the fact that most of that $500 winds up in the pockets of a California company rather than in China itself.

When you take into account the value of intellectual property around the world, it explains why some of the most valuable corporations to be created in the past 50 years are American, despite supposedly persistent trade deficits for the entire period of time. Because we're surprisingly good at creating new intellectual property--thanks to a legal and cultural environment which encourages greater levels of risk-taking than does Europe or Asia.

So all this finger pointing over trade deficits, all the actions taken by the Trump Administration, all the hand-wringing over how the US is somehow being 'bled dry'--is all based on a faulty metric

And so long as we keep measuring the wrong thing, and talking about the wrong thing, we'll keep doing the wrong thing to fix the wrong problem.

If this works...

There's a lot there to hang your hat on, especially for something not slated for completion until 2030.

When a government--any government--announces they are doing something for a stated reason, I tend not to believe the stated reason. In this case, if what they wanted was to guarantee a steady flow of electricity by creating a supply of 'on-demand' energy sources that can ramp up when there are disruptions in the grid due to weather, they'd be looking at things like natural gas turbines (basically jet engines attached to power generators which can spin up and down at a moment's notice).

Not nuclear.

So I'd be looking for an underlying explanation that makes more sense than "we need an on-demand source of power, so we're going with the one energy source that does the shittiest job with on-demand energy supply."

Unless there has been a breakthrough in on-demand nuclear that I haven't seen...

I mean, wasn't that the whole point of the Euro being rolled out around 1990, to allow Europe to dominate (or at least counter) US dominance in the financial markets, and unwind the effects of Bretton Woods?

And how is that going?

How many folks here have been in the situation of trying to figure out what the hell marketing just sold, so they can build something that matches?