There's this really amazing concept you're going to learn about one day, AC, called a proxy. It turns out that in medicine, very often, you can measure a proxy to give you an indication about an underlying biological process. The proxy is not as good as measuring directly, but is better than not measuring at all.

Here's some other examples of proxies: all diagnostic signs and symptoms of Alzheimer's on living patients (gold standard Dx requires post-mortem histopathological review of brain tissue); heart failure, where we cannot measure neurohormonal dysregulation that actually causes the issue directly; and ironically, hypertension itself, where blood pressure readings at the arm are themselves a proxy for vascular resistance and cardiac output within the entire circulatory system, because noone is getting central aortic pressure routinely measured.

I hate the stupidity of your comment. The use of ML to discover new pattersn that act as good proxies for disease has been, and will continue to be, hugely beneficial for human health, reducing undiagnosed chronic disease and prompting interventions that will allow people to live healthier, longer lives. And then sneering little know-nothing twerps like you come along and think you have discovered some reason it wouldn't work, based on nothing but misplaced overconfidence in your own abilities and a predetermination that the biomedical scientists, engineers and clinicians who have worked on this were both stupid and venal.

As I said, you’re conflating two issues. I don’t like RTO mandates either! But that’s no reason to claim that trains are inherently uncomfortable or unsafe, and I’m glad you’ve backed away from that absolutist statement

That already exists! Teams Copilot will do exactly that for you today

Agree Gurman is worse, but I looked up Kuo's track record and 70% is what he manages, ie little better than a coin flip.

You're conflating the downsides of not allowing remote working with the perceived downsides of enabling public transport. That's.... muddled. Public transport provides plenty of value beyond commutes, and the future of work and of commuting is mixed modal anyway: people working at home some times and outside the home at other times, and getting around on foot, on bike, on scooter, on buses, trains, metros, light rail, heavy rail, boats, and sometimes private cars too.

The existence of trains does not cause the existence of RTO policies, and trains can be very comfortable and safe, and it's just weird and deeply, deeply American, to claim that they can't.

Am I the only person to recognise his name and know that he has a decidedly mixed record when it comes to predicting future Apple products? He gets it right about 70% of the time, so a little better than a coin toss, but not something I’d use for making money.

One of the most important reasons that many people in European cities don't have cars is that there is great public transport in European cities. For the 20+ years I lived in Hampstead, St Johns Wood, South Hampstead, West Hampstead and Kilburn, I didn't need a car because I had such easy access to multiple tube stations, train stations and buses. Now I live a little further out and while I still have easy access to tube and bus, I do have a car, although we don't use it that often, because there's some journeys that are easier in a car. But I'll happily use tube/train/bus etc. This kind of mixed modal use of transport is super-common in Europe

You have not got a clearly resolved position on this topic.

This is just a trope you guys tell yourselves. There’s no evidence basis for it. Where you have public transit, it’s heavily used, even if you won’t do it. You sound like Thatcher about people taking buses (it’s a notorious quote, look it up)

It’s also not true that these reports are more often revised down than up. Instead, they are revised down a bit more often in periods of contraction, up a bit more often in periods of growth, and revised less and in no particular direction in periods of stability

It doesn’t quite work like that. Some users need to prioritise timeliness over accuracy (eg traders and investors) and others need vice versa (eg academics). The former would rather act fast and live with the risk of acting on somewhat inaccurate data. The Fed, for example, cannot afford to wait two months for revisions before making interest rate decisions. Their choice is decide on the basis of incomplete data or decide on the basis of no data.

Here’s a good program about this

https://www.bbc.co.uk/sounds/p...

Possibly, but unlikely. The discrepancy would have shown up sooner. The most obvious explanation for the discrepancy is that the economy is weakening faster than the CES birth-death model can respond.

It’s typically going to be better than buses, due to scale economies and lower rolling resistance. Depends on load factors etc, and EV buses have cut the gap, but steel wheels on rails vs rubber on roads is a big advantage

That’s a direct result of the discouragement of investment in trains in the US. Elsewhere in the world, a train is a comfortable and safe experience.

But you need to know enough to know whether to be concerned by a 25 to 50% revision. For example, you’d need to have detailed expert insight into the CES birth-death model and how it responds to fast economic contractions to come to an informed view of whether the downward revisions of May and June this year were implausible or not. And you’d have to be able to understand the cause, and rule out the more likely cause, which experts have said is that the early estimates have been consistently too *optimistic* for many months.

The much more plausible reason for somewhat more frequent downward revisions of the CES data since 2023 is that the economy is contracting faster than the BLS’s model can adapt to in real time, especially in the small-firm sector. Consequently, downward revisions (due to the birth-death model) are entirely expected. This doesn’t indicate a flaw in the BLS process per se — just a lag in modeling adjustment during a turning point in the business cycle.

The rapid economic contraction since 2023 has meant the BLS birth-death model is overestimating the number of new businesses, and underestimating closures. This causes *early* CES estimates to be biased *upward* — the complete opposite of what you think is happening.
Later, when more complete data arrives (e.g., through late reports or benchmarking to unemployment insurance records), the overestimation gets corrected — resulting in downward revisions.