FOSS folks: What will it take to unseat The Cloud as the arbiter of the software we create and the only way to make a living off it? How can we revisit the way we do our community-level, volunteer work so that instead of feeding the parasites, we feed those in need?
Discussion
Loading...
Post
@zkat The tech world has to re-focus on personal computing, and the idea that our computers need to have integrity – be distinct from the network services – first and foremost.
One of the biggest contributors to cloud over-reach is that there is no visual language that communicates what runs locally and what runs in the cloud. Most people have no idea as they're not receiving any cues about local/remote status.
Another factor is the rise of special-purpose gadgets that are networked. They are not general-purpose and have no OS UI for the user to interrogate; they are just opaque. A product line that starts out as autonomous can switch to cloud-based without the user being aware.
@zkat get really amazingly great at designing and writing useful local-first, offline-first tools, with proper multi-client zero-configuration peer-to-peer abilities for when it makes sense and a mathematically proven correct BASE consistency model fit for purpose ¯\_(ツ)_/¯
Tools which are absolutely *nothing* like any commercial cloud products (nor monstrosities like unlimited forever-growing write-only blockchain cancers)
@zkat Offer an alternative
Cloud first → Local first
Always online → Offline first
Client/Server → Peer-to-peer
Centralised → Distributed
ACID → BASE
Not in anyway structured like a corporate hierarchy
"Making money of it"
Clearly nothing like corporate money
Not saying no profits, but not the ongoing exuberant profits of rent seeking cloud racketeers
But I know there is a niche for tools like these made by individuals. Humane tools made by humans for humans.
@zkat Ive wondered about this a lot, too. I have an idea for neighborhood tool library software that i’d also like to host locally. But I honestly have no idea how to run a server out of my home or if thats even a good idea.
Which is to say, I cant answer your question but I sure hope someone else can. I like the idea of an easy way to build a sort of “datacenter”, made up of a few computers distributed across some neighborhood homes, that run community applications.
@zkat The basic problem is that the skills to create the software do not overlap with the skills to package and sell it, so until we can partner with sell it people who don't simply pillage, that's unlikely to happen outside of a few small cases.
@zkat Lean into the strengths of local-first, in domains that cloud can't compete in:
Tight integration with desktop & user.
Native UI.
Lightning-fast latency by relying on only local state.
Close integration with local user data without hurting privacy.
Long-term stability and reliability.
Backend agnosticism.
Customisability, hackability, introspection.
@zkat the path of least resistance will ALWAYS win. Make it easier to do the right thing. As easy as possible.
@codinghorror this does not answer my question. It is literally just rephrasing the same question I asked, from a different angle.
WHAT do you do to reduce that resistance?
@zkat correct question but too complex to answer through a straw — a blog post about this would be incredible and so helpful to so many.
@codinghorror
Please ping us here if you do write that blog post. In principle I wholeheartedly agree, generally phrasing this as "Convenience Matters" (https://one.mikro2nd.net/pages/convenience-matters/ - my own blogpost on paths of least resistance, but not helpful to the question at hand: How do we get people off the "Cloud" religion?)
@zkat how do we usefully talk about why The Cloud is so useful from an infrastructure management POV, as well. It's hard to say "bring servers back to the not-cloud" when we don't really have easy ways to talk about how hard it is to just run kit somewhere.
@aurynn I mean we had a lot of solutions for this before, and they were still in development until The Cloud started making it more convenient on some key metrics. I think we can do better. And I think a lot of the Con is making people think that YOU need a cloud provider, that YOU need to scale, anywhere to the same extent as the companies that DO benefit unmistakably from "The Cloud"
@zkat I don't need to scale as much as I need someone else to think about resilience of the hardware and having a supply of spares on hand and going to the datacentre?
@aurynn I think we also need to address realistic resilience needs.
You, as a small operator, don't need the level of network resilience/redundancy as a Fortune 500 company that will lose millions of $ per minute of down time.
Not that redundancy is bad, but that you can go as far as having a simple two-small-machine setup and that will probably give you all the uptime you'll ever need.
Likewise, writing software in hardware-hungry languages is a luxury we've afforded ourselves because of the convenience of the cloud. Driving our software towards embedded queues, sqlite, systems software developed by communities for their own use, that can run on $30 hardware in a closet is very different from needing to spend hundreds of $ on each machine.
@zkat @aurynn the thing cloud gives you as a small operator, is actually the ability to scale down, not up. ramping up doesn't really happen, so that's not a huge concern. but spinning up an entire machine, and all the attendant infrastructure, for something which could run on a tenth of a vcpu, is also tremendously wasteful. the only way to get that kind of solution in the past was to co-locate lots of websites on physical hardware, and have somebody manage. dunno if i want to bring back cpanel
@dotstdy @aurynn what if the hardware itself were so cheap and commodified that for less than US$30-80 you had a complete solution that could scale as much as you ever needed, but didn't waste a ton of resources, and even let you add redundancy?
Currently, the smallest cloud offering/droplet you can get is usually around $4/mo, with 256mb of RAM and a single core. $30 here can get you an ARM quad core with okay-enough perf and around 1gb of RAM. Not HUGELY wasteful, but honestly? Aside from going fully embedded (which complicates dev significantly), you can't really go cheaper than that (I've looked).
1+ more replies (not shown)
@zkat 2¢: It has to be less niche at a broad cultural level. It has to be something folks outside of tech, tech-adjacent, and burnt-out former tech folks care deeply about.
The greatest trick the Cloud ever pulled was making dissent boring as hell.
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled