That does not account for survivorship bias. It makes no mention of compressors from the 30s and 40s that were universally terrible, so none existed for long enough to evaluate their MTBF.
It also does not evaluate reliability of units from the 50s through the 2000s, which is apparently when everything became unreliable junk.
In order to properly control for survivorship bias, you need to tabulate reliability figures for each year (and for each price point). The handwaving assertion about manufacturing consistency does not do that in the slightest.
>You're making an unnecessary distinction between intra-brand survivorship bias and inter-brand survivorship bias.
No, I am not. A $10k fridge from a premium manufacturer that makes no budget models will last decades. A $1k fridge from a budget manufacturer that produces no premium fridges will last a few years.
There are very few manufacturers that make models spanning the entire reliability (and thus price) gamut.
>Does any party have access to these data, realistically?
Probably not. Which is why we cannot assert with any solid evidence that appliance reliability has declined over the years—we have no way of conclusively ruling out survivorship bias.
>There are very few manufacturers that make models spanning the... gamut.
How does that relate to survivorship bias?
>Probably not. Which is why we cannot assert with any solid evidence...
...anything.
If you crank the standards for "solid" high enough, you can make literally anything become unknowable.
You can always ask for better evidence, in any situation! Historically this has been exploited in denial campaigns (tobacco, climate change, etc), so there should be some wariness surrounding this rhetoric.
That said, N=1 here. Let's not make more (or less) of it than there is.
>we have no way of conclusively ruling out survivorship bias.
Is the corollary that we have no way of detecting planned obsolescence?
I hope I'm misunderstanding. If not, it's the fox guarding the henhouse.
You said that inter vs. intra brand survivorship bias didn’t matter. I think it does, since the former likely has a much bigger effect (models from premium brands generally last longer overall due to higher reliability, whereas random units from cheaper brands last longer due to pure chance).
>...anything.
>If you crank the standards for "solid" high enough, you can make literally anything become unknowable.
This is arguing in bad faith. There is a high level of evidence required to support the extraordinary claim that modern appliances are generally less reliable than old appliances, after controlling for price point. I believe a reasonable amount of evidence to support this claim would require unbiased sampling of a statistically significant number of appliance service records over the years. Otherwise, we’re just idly conjecturing.
That said, I take back my claim that the requisite evidence is unobtainable. We unequivocally know that cars have gotten much more reliable over the years, based on purchase frequency and service records (remember, cars’ odometers used to top out at 99999 miles because they weren’t expected to last longer). There is no reason we couldn’t do similar surveys with appliances.
>Is the corollary that we have no way of detecting planned obsolescence?
No, the corollary is that it would be equally difficult to infer if planned obsolescence has gotten more frequent over the years. That’s very different from being unable to detect planned obsolescence at all. It’s easy to inspect a particular appliance (or line of appliances) and find evidence of premature failure points, as the linked article nicely did.
It also does not evaluate reliability of units from the 50s through the 2000s, which is apparently when everything became unreliable junk.
In order to properly control for survivorship bias, you need to tabulate reliability figures for each year (and for each price point). The handwaving assertion about manufacturing consistency does not do that in the slightest.