Whenever people talk about tariffs on Canadian goods I just remind them of the last time that happened. We slapped 'em back with similar tariffs and suddenly the US tariffs were lifted. https://www.canada.ca/en/department-finance/programs/interna... Not sure why buddy even talks about this because we have no problem doing the same thing.
> According to Deepseek, the tariffs could shave as “little” as 1.7% or as much as 8% off of Canada’s GDP, with the baseline estimate around 4%. That’s roughly in line with the Bank of Canada’s projection.
The BoC estimates 6%. I guess it is accurate to say the 1.7% - 8% estimated by Deepseek is “roughly in line”, but it is such a large range as to be pretty useless. Though, if you look at the BoC’s ToTEM model for GDP, it is pretty useless as well, relying on shaky assumptions like a regression to the mean over 6 quarters, which isn’t accurate when it matters most.
GDP numbers with decimal points need to be taken with a grain of salt.
The US economy will be hurt as well. Housing costs without Canadian lumber are going to skyrocket, especially with the different natural disasters and home building.
The interesting thing to analyze is the impact of the deal.
The retaliatory tariffs from Canada would be really painful and impactful to the US. The auto industry would implode. The dopey magas in the Midwest would be all laid off in weeks.
Unless we’re going full fascist and rolling tanks, this is positioning for some sort of deal. They cowed Columbia with this stuff, there’s a limited time to play the playbook.
> shaky assumptions like a regression to the mean over 6 quarters, which isn’t accurate when it matters most
For most second- or third-order effects, this isn’t a bad assumption. When you’re putting a 25% tariff on your largest trading partner into an economic model, you’re already in EVT territory—the central tendency is less relevant than the ranges.
At the scale of a national economy it's just about impossible to comprehend what a small digit % even means (I have some luck reimagining everything as 1 in X, which tends to feel more relatable at least for human centric topics).
Nothing made me feel as dumb as reading its chain of thought. It made it click for me how hopeless we will be soon, once we cannot comprehend even the reasoning for some things.
I'm not so sure. LLMs are more like human simulators and are pulling this "reasoning" from what they've seen in the training data. It therefore might be possible that they cannot be more intelligent than those who wrote the source material.
This is not a wrong sentiment. I'd be super interested in hearing an analysis of its CoT. To my ears it was a nearly schizophrenic roundabout with a lot of contextual things like "that's high" but I don't know why it thinks that.
What I think is incredible about this example is how adept it is at mixing precise calculations with rough heuristics. That's exactly what I was taught, to validate the numbers by asking "Does this answer seem reasonable?"
ChatGPT doesn't do that at all, as far as I'm aware
“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”
"The model predicts a 15.5% GDP contraction under simplified assumptions. Real-world impacts would likely be milder due to exchange rate adjustments, policy responses, and import reductions. This highlights the sensitivity of trade-dependent economies to tariffs and the importance of holistic economic modeling."
And what about the impact on employment?:
"A 15.5% GDP contraction could lead to 8–12 million job losses in Mexico, disproportionately hitting export manufacturing and informal workers. Unemployment rates could surge to 25–30% (including underemployment), with long-term scarring effects on labor productivity and social stability. However, aggressive fiscal/monetary policy, peso depreciation (boosting export competitiveness), and diversification to non-U.S. markets (e.g., EU, Asia) could soften the blow."
I closed the tab and flagged the item after the third paragraph, I mean, FFS, it's the sort of stuff that gets downvoted into invisibility when people do it in HN comments
This should be not just flagged but the entire blog blacklisted from HN. For fuck's sake, there's literally no value in a post that's nothing but "here's what some AI said about it."
Would be interesting to get an AI take on the other side, what is the most likely range that the US stands to gain from this action?
After that, why don’t we just have these two AIs game theory it out with each other? They seem to have a more objective and holistic incentive structure than former TV celebrities and career politicians.
I agree its a terrible idea, but an economics text book will tell you that a tax on a transaction will be split between buyers and sellers based on the elasticity of demand. So its not quite that simple. But imo not a good idea either.
Those domestic producers depend on materials and components that are imported. Domestic production will also be highly negatively impacted by foxes increasing on everything from computers to steel to truck parts.
All this says is that a LLM model can roughly reproduce the model & results that a human did; they can both produce logical sounding bullshit about a incredibly rare event for which they don't really know the answer. Logical reasoning for the actions of an unpredictable idiot would be laughable if it was not so serious.