Biggest problem is a company like eight sleep has the marketing. So if someone really wants a temperature controlled bed, it's hard to know what a credibly good one is. I *think* Chilipad is a good one, but it's a pretty pricey thing to evaluate and thanks to internet-everything, it's not like you can see for yourself.

But yeah, Eight sleep deserves every amount of bad press they can get for being such a douche company.

They have local controls on most recent models, *however* the controls will deactivate unless the cloud control has blessed the user in the last 24 hours. Before getting going, they'll talk to local phones without internet, but *only* for the end of getting the WIFI set up. They know exactly how to make local phone control work.

It was never about cost savings, it was always about a path to forced recurring revenue. They opened with early adopters not having to pay subscription fee, but still forcing them through the servers. Early adopters also didn't have to pay too much and actually had a decent warranty. When they managed some good momentum, they cranked up the price, tanked the warranty, and forced subscription.

The code was written toward the purpose of forcing the users into a monthly subscription.

The goal was not to deliver the best user experience. To the extent they have tried to accommodate demands for local control, they have predicated it on having relatively recently been 'blessed' to let the user do that within the last few hours. That takes explicit effort to implement a local control loop and make sure it gets approval.

My wife insisted on it and we bought one when they were getting started and relatively cheap and the subscription was not yet required. We've been grandfathered in so we don't pay the subscription, and thanks to the leaks we have been upgraded to the latest model, so I have familiarity.

They are a shit company with a decent hardware design (now) that stops short of being good all around precisely to gouge users.

Even worse, they have a local control loop, but they deliberately cripple it.

If the bed is 'on' (which is only allowed through their cloud connection), then you can locally adjust things fine. However it will refuse to do this if the internet hasn't approved the device to operate locally.

This 'enhancement' was added after people demanded local remote or buttons or *anything*. They implemented an earbud-style tap side of bed N number of times for adjusting temp or dismissing alarm.

So they know precisely what they are doing, it's not dumb engineering, it's malicious engineering.

But what if no one is ever going to use the output anyway? Might not need to check it.

I've dealt with *way* too many business processes that have people generate obscene amounts of prose that no one will ever read or even skim or reference.

I remember one of these companies championed that they used LLM to complete an important 'overhaul' of their source code. The 'overhaul' was generating separate document detailing all these uncommented functions and what the LLM guessed they were supposed to do and how in plain text. The theory was that if one day they actually wanted to start porting this code to something else, that document would be 'helpful'. And of course:
- They never will do that porting
- Even if they did, the developers will likely ignore that document.

Question to what extent was revenue reduced versus deferred. If 90% of their customers couldn't reach competitors either, was revenue lost or did it just happen later?

The thing is that this is terrible for all the outages to be aligned for the internet users, but for the providers, the thought that outages are likely to align with competitor outages might be a pretty solid mitigation, so long as the outage doesn't exceed what they might incur themselves. Even a longer outage common with competitors may be better than a shorter outage that *only* impacts them.

Certainly if AWS was down enough that someone could get a competitive advantage by moving it would drag on them, but if it's not *much* worse than their own outages, well there's a comfort in making sure your competitors are more likely to go down with you.

But what is the tax? These services are down and no one is blaming the companies, they just say it's an Amazon problem. No one is penalizing a company because their Internet services went down just like most of their competitors

Think the point is going beyond external. If you are converting to hardwire, I'm picturing removing the power connector and putting some screw terminals down.

A barrel connector is going to be a couple of rather large solder points. A USB-C connector is... not going to be that.

As I recall, you can do wireless charging, and even exclusively so if you want no charging port at all. Also, plenty of devices still do non-USB-c power (e.g. if it takes AC power in, that's not required to do USB-C, and if it's a car, then it's CCS2).

There's no sign of a successor to USB-C form factor in the space that EU mandates its use. IIRC, they even wrote the standard to leave some wiggle room to adopt such a successor should it arrive, but the industry seems to have settled into USB-C as an DC power strategy up to 240W with no interest in others.

Barrel connectors can be like $0.50 cheaper, so a fair number of cheap devices could balk reasonably at how USB-C drives their prices higher and they don't need USB-C. But a more 'advanced' connector is not in the cards.

Obviously the next step is CCS2 over USB-C, duh.

It's GNU/Hurd or nothing

I think any dramatic change from how you currently run things to a different way is full of risk. Just because it's Linux doesn't really do much in the face of who knows how much hard coded this or that they accumulated in their infrastructure management.

People's infrastructure management tends to be ugly and locked in to how they do it in various ways.

Azure may be utterly capable, but any difference is a huge headache, particularly the longer the 'old ways' went on and how many people along the way left the company.

It was widely rumored that in 1998 Microsoft tried to force Hotmail to use Microsoft infrastructure and met with predictably miserable results. Hotmail was more about trying to show off their infrastructure products that as an offering in and of itself.

Microsoft might be a bit more conflicted on github, but clearly that sentiment persists.

Unfortunately looks like they are sufficiently impatient they've pivoted to trying to fire up more coal instead...

Most bubbles are rooted in something real, the dotcom, the housing market. You do have things like crypto that is more purely fictional, but most mainstream bubbles are rooted in real stuff..