What we need is a cloud platform that allows people to deploy code to computers on their home network, or any cloud, or any data center... one that encourages and assists with cloud native design and aims to reduce load rather than hope you architect incorrectly and happily charge for the waste.
Perhaps I’m being naive, but it feels like a tooling problem, not a social or political or economic one. Anyways, I’ve quit my job at Stripe and just finished YCS19 with the goal of closing the gap between Facebook/Mastadon, Medium/yourBlog, Minecraft.com and a docker image...
And you’re totally correct about renting home server “boxes” too - except most house holds already have plenty of compute and connectivity - it’s just that installing / managing server software is completely out of reach to most people (and some developers!). It doesn’t need to be grandmas running router software - just the family nerd able to create on the internet again - and reliable enough software you’d actually invite grandma to signup!
Sandstorm tried this, and they got very very far.
I think it was ingenious, but the sysadmin in me was very skeptical about the HA/scaling story (there was none). I guess it's fine for self-hosting for yourself&friends, but there should be an easy way forward - after all, Kubernetes commoditised clustering (I mean - once you're past setup).
Where can I follow your efforts? You mention KubeSail in your profile, but that seems developer-oriented?
> We probably should have spent our time making the platform work better for developers and end users instead [...].
I keep a number of folders / bookmarks / stickies, where I save all the quotes like this one, so that I can then stumble upon them randomly. This is the single most important thing to keep in mind when building any kind of a platform. Developers particularly, are often the most important users of your product, since they will over time help drive the momentum and keep on bringing value.
Thank you for this reminder.
Ease of deployment is one thing, but managing and troubleshooting these deployments is another. Ordinary people generally don't want to do any of these things.
There are some people who have both the technical aptitude and desire to do these things, but they'll likely always be a small minority. An even smaller minority will be willing to do these things long-term without getting paid.
I've been running my own servers from home for a very, very long time and it has never been an issue.
Running a social media node would probably count as "intended for the general public", but I'd bet that Comcast wouldn't really notice or care unless the amount of traffic exceeds a certain level or they start getting abuse complaints.
Here's the twist: the company (presumably a non-profit or social purpose corp) would also offer the ability to buy your box, at which point they mail it to you; when you plug it in, your existing data is copied to it and you begin to self-host instead.
This makes it easy to sign up for and try out, while also offering a path toward self-hosting, once you decide you like it and don't want to keep paying the monthly fee.
This is something that the SoLiD (Social Linked Data) initiative, based on web and linked-data standards, is expressly built to support; your "box" would implement what the initiative calls a Personal Online Datastore, or POD. Though "federated" third-party hosting of such datastores is also allowed for, of course.
You could rent the box or buy outright with money to develop paid for by a premium on the box or by the rent wherein even a modest rent would outstrip the cost of buying like a cable box.