AFAIK, variable interest rate mortgages only require payment increases if rates rise high enough to no longer cover interest payments.

Anyway, I hope you’re right. As confident as I am in a pending financial disaster, it would be great for everyone if that doesn’t happen!

He also did $hundreds of labour to convert the kit.

Not to detract from what he did—it’s genuinely awesome, but it’s not practical at scale.

I know that, yes. But it’s relevant to my point: AI costs more to train and run than its value. That Microsoft, Google, Meta, OpenAI, Anthropic, SpaceX and hundreds of smaller players are all losing money at the same game just makes the problem worse.

There is no path to profitability for AI.

The cashflows needed to pay back investment to date is already impossible, and it’s scheduled to accelerate.

And Microsoft being tied to OpenAI is largely about continuing their “growth story” to justify their valuation. The “value” of their ownership stake, even as huge as it is, is largely irrelevant; once the house comes tumbling down, it’s going to be a very long path to recovery for anyone involved in AI. They will write off many billions and be responsible for many times that in evaluated paper wealth.

Google, Meta, and Microsoft will all be fine, eventually. They have enough stable revenue streams outside of AI to weather the storm, most likely (I’m not sure how leveraged they are, but with string cashflows, I imagine they’ll survive.)

Musk’s pile of companies are dicey. OpenAI and Anthropic won’t meaningfully exist in 5 years.

So, I’m pretty confident that a variable rate mortgage is the right choice right now. But even if you aren’t as doom and gloom, it seems pretty clear that signing a 5-year fixed right now is unlikely to be a good idea.

I didn’t address your point about housing. You’re right that low interest rates have exacerbated the problem, but the bigger upstream causes are more that:

1. It’s political suicide to run on a platform of reducing housing prices. People who vote are, disproportionately, homeowners who don’t want numbers to go down.
2. The missing middle. We’re building the wrong kind of housing, and zoning bylaws are preventing unit densification where people want to live.
3. The lack of non-market-rate housing options. Affordable housing options have a disproportionate effect on housing costs. Canada hasn’t invested in, basically, any affordable housing since the 70s, and we have some of the lowest non-market-housing rates in the developed world.
4. Housing being held as investments first, and housing second.

To be affordable, housing needs to drop 50%. Interest rates being low contribute, on the margin, but don’t account for that big of a difference. You’ll notice that housing prices barely dropped at all, and even continued to rise in some areas, when rates skyrocketed 4 years ago.

What more than likely happens instead or also is the survivors get a lot bigger than they are.

No. No, they won’t. They’re all losing money hand-over-fist. And they’re all circularly invested with each other. Microsoft owns OpenAI, and OpenAI uses Microsoft compute which they pay for with ownership. Microsoft valuation is inflated because of OpenAI, and when the bubble bursts, they’re going to lose a lot of money.

If you’re interested in economics/finance, I suggest reading about the circular cashflows and investment in the AI space.

To make it even worse, they’re effectively lying in their accounting, claiming the compute they’re giving away as a “marketing expense” instead of operating costs, while booking the value as revenue. The profit : cost ratio is even worse than headline numbers suggest. They are burning through multiples of all of their revenue on operating costs, before they even start to pay for expansion.

In other words, it’s fucked.

Sure, AI is here to stay—small to midsize models are already trained and cheap to run. But it’s also not going to make significant profit for anyone.

No, it’s not. Housing doesn’t require much to sustain value, while AI valuations require the market to continue to believe it is a “growth industry”, which means it will have explosive, exponential profit growth in the future. That is not possible, and the longer the market remains irrational, the more expensive it becomes.

Housing is, literally, unique. Every home is brought to market independently, and homeowners are unwilling to accept a lower price, so they sit on inventory.

There is also a lot of political pressure from homeowners to sustain valuations, especially from aging homeowners who are depending on the home valuations to fund retirement. And this is the demographic that votes.

vs. AI: in the coming months, when OpenAI, Anthropic, or any other smaller player needs to pay $billions to a supplier or default, and they can’t raise more capital, they instantly go bankrupt. Then the whole house of cards comes tumbling down.

I wouldn’t do that, if you care. See my comment above for my reasoning. Variable is the right choice right now, imho.

FWIW, I think variable was the right choice for you. My term is up soon, and we’re going variable.

Over investment in AI is unsustainable. There’s no path to profitability for the industry. There is going to be a major market correction soon—and it must be soon because the increasing costs of investment are burning through capital at an unsustainable rate.

This will directly put downward pressure on rates, as liquidity rapidly contracts as the circular investments in AI suddenly collapse.

And also indirectly, as a major driver of BoC rates is inflation, and one driver of inflation is chip prices exploding due to AI.

And also indirectly from stock market wealth losses: when investment valuations collapse, that reduces wealth, which in turn puts downward pressure on the housing/construction/materials markets (b/c affordability drops) and, hence, inflationary pressure from those industries.

And then, even more indirectly, the US will be in a politically untenable situation wrt their very expensive war of aggression triggering energy price spikes during a crash and recession. They will be forced to back down, which should stabilize energy prices.

So, I think you’ll do very well on variable, even if it rises a bit in the short term.

That doesn’t pass the sniff test. His party had 2 more years in power, which is plenty of time to choose a new leader ahead of the next election. If he wanted out, he could have stepped down.

No, I think this was the NDP senior strategists/leadership expectations that the opposition was in such disarray that they could cruise to an easy majority. We’re they not paying attention last election? BCU folded as soon as it was obvious they were unelectable; it wasn’t 4D chess to think the nascent right-wing parties would fold even earlier.

They also handed the BC Cons an excuse for not having an election platform—they were in disarray just trying to respond to their leadership turnover and the unfair election, so just saying “trust us, we’re better” seemed reasonable to voters.

Its also incredibly hard to catch erors in you own writing. Skilled writres are usually good readers, and we have mnay silkls to dcpoyher txet, often with out noticeing.

(Errors introduced deliberately to illustrate.)

There’s a reason virtually all books make it to print with errors. It’s very hard to catch all errors.

It’s also a fantastic use case for LLMs, since the structure of language is the exact type of low-hanging task LLMs are actually good at, even with cheap/fast models.

It was. I was intentionally trying to sound like an LLM, using some of the most common phrases and sentence structures. :)

I just watched the whole thing, and I don’t recall Vancouver being mentioned. He’s based in Peel Region (Ontario), but he interviewed people from many jurisdictions, including California, and someone from the EFF.

It for sure is. The storytelling style with complete lack of personal voice is jarring. Using cliches matter. The Uncanny Valley of text that sounds human—but clearly is not—builds to an unavoidable conclusion:

The article is AI slop.

Still, I did find the content interesting, but it’s a shame the original source wasn’t linked instead of a slop summary.

Except they’re not? Very few tools are deterministic. Only Mathematics is, and that’s only a small subset of tools.

You’re free to argue that non-deterministic tools lack value, but that’s clearly false: OCR only exists because of lossy training machine learning algorithms. Or, going even more basic, Monte Carlo simulations are by far the most efficient method of testing entire categories of problems that are intractable deterministically.

LLMs can be shitty, as a business practice, due to their unpredictability, but (hopefully obviously) that depends on how it’s used.

You’re correct but still completely missing the point.

The only way to implement anti-LLM rules is to detect AI-assisted code. That, in and of itself, is a problem because it’s impossible to detect good code that was AI assisted. And you don’t need rules against slop because it’s bad code.

If I’m following it correctly, Linus is advocating for tightening submission requirements before it gets to human review, which directly targets the harms FOSS projects are having under the deluge of slop: it raises the bar so that the slop needs to be good enough to warrant a review, which filters out must of the noise, leaving more of the signal.

Anti-LLM policies are like having closed-book take-home exams: you’re left wildly guessing what’s cheating (or not). Instead, make the exam open book, but expect better answers.

Linus also says AI is a useful tool, and he is an expert, so we must accept that he feels that way. I don’t know if he’s commented on the environmental/social/political consequences of AI hyperscalers, but that’s a separate argument.

The tech is here. It’s impossible to detect. It’s foolish to think you can ban it’s use

And it can be run locally, so there’s no stopping it, even if there was political will to do so. (And with a reasonable economic and ecological cost, no less—ignoring IP/political consequences here, since that’s a whole other topic.)

I can’t wait! I’m really going to notice a 0.1s faster restore time.

My DBA friend builds a lot of sets for school productions and community theatre. It’s only a matter of time…

When you’re vibe coding, you need to narrate long prompts, including things like “And remember not to hallucinate or make any errors.” All that is a lot faster to say than type.

Clearly, you’re not vibe coding, and your job will be (checks notes) automated away by someone with a #grindset.

/s, in case that wasn’t obvious.