This is probably for an audience less enamoured with the Pi than the HN crowd. Someone that's more interested in getting to a working result than having to yak shave for a couple days or more to do the same.

For someone who doesn't have a Linux background, "just put it on a Raspberry Pi" is kind of like saying "You write a distributed map reduce function in Erlang". Ie: it's easy if they know it, but if they don't then that "just" is doing a lot of work there.

Pre-installed is almost certainly the way to go for such a person.

When it comes to Home Assistant, the Pi is actually a much more pragmatic option.

It works out of the box, is very easy to source (hell some brick & mortar stores sell them), has very good Linux support due to its popularity, and makes up a large part of the install base meaning HA support for it is unlikely to get deprecated.

Right, but the fact that running it on a Pi with Linux is the "much more pragmatic solution" is already ruling out about 90% of the US.

Keep in mind that Home Assistant provides ready-made images that behave like an appliance and can auto-update. In fact, it doesn't even give you a shell/SSH by default and such access is discouraged.

Thus the Linux/RaspberryPi underlying complexity is irrelevant to the user - the "complexity" is to dd/BalenaEtcher/etc a downloaded file to an SD card, put the card in the Pi and connect it to power. From there it's available over the network and can be configured through a web browser.

> is very easy to source

I was with you until this point. A Pi hasn't been easy to source for almost 4 years now.

I purchased a home assistant yellow and my experience was anything but “ready to use”.

You have to build the damn thing, which isn’t hard per se since it’s ultimately only 3 actual components, but it still took me some time and felt complicated since it involves attaching a heat sink with thermal compound on a CPU.

And then the software install process isn’t totally amazing either since it involves flashing a USB stick, but also needing to choose a few very non-obvious options.

Should I install HA on the EMMC and later move my data-disk to the nVME drive or install the OS on the nVME drive directly? Google random forums to find out what people think of this decision first I guess.

I mean I think it’s still a good product, don’t get me wrong, but it is still very much a power user thing.

Which is probably fine because setting up HA itself when you have an install isn’t exactly a picnic either.

I don't know if it's changed but I felt like a super genius a few years ago when I finally got my HA up and running on a pi, and I'm a linux person and former system admin. There's Home Assistant and Home Assistant Core, Docker or not Docker, install HACS or don't. Some things don't seem to work unless you're on a Docker container but then it's a pain to ssh in and find folders to install stuff. I really hope it's better now as my HA install has mysteriously died and I haven't had the heart to dig in and see what the issue is so I'm guessing I'm going to have to start from scratch.

> HA in particular makes it very easy to chose how to install

This is the problem with lots of stuff similar to HA when it tries to break into a non-enthusiast audience: people don't WANT to choose how to install it. Most of the time they have no clue why they would choose one thing over another and giving them those choices is confusing and overwhelming.

It's like starting a an intro to Nix tutorial with by asking if the user wants to enable flakes.

I say this as a very active user of HA & Nix for 5+ years.

With HA it doesn't help that their installation docs are a mess with solutions that don't provide the same features.

I've had HA for +4-5 years too.

HA is similar to self-managed Kubernetes: easy to install, a bitch to maintain. Updates seem to constantly break services and configurations.