💾 Archived View for jojolepro.com › blog › 2022-12-12_forever_computer › index.gmi captured on 2023-06-16 at 16:11:13. Gemini links have been rewritten to link to archived content
View Raw
More Information
⬅️ Previous capture (2023-01-29)
-=-=-=-=-=-=-
Jojolepro
Blog
Projects
Music
Quotes
GitHub
Archives
re: Forever Computer
Ploum started an interesting conversation on what a computer that would last
a very long amount of time would look like.
gemini://ploum.be/2022-12-03-reinventing-how-we-use-computers.gmi
I have had some thoughts related to this too in the past and did some
experimenting with kiss linux to see how far we can push minimalism before
it becomes too limiting.
I have also noticed a trend which might be applicable to humanity in general,
which is that we almost always try to move forward: do more, faster, easier.
People focusing on security and stability, those who take less risks, will be
left behind.
The question is: are you fine being left behind for some things?
Are you fine doing plain old C programming when people are using rust and code
generation tools? Are you fine being constrained to an environment that will
feel slower and slower *compared to the alternatives*?
For this blog post, I propose the following method:
- Define the properties of a forever computer.
- List out all my goals and for each determine if they can be done on a forever
computer or if an alternative is needed.
- Come up with a list of hardware that fits the properties and remaining goals.
Forever Computer Properties
- Can last 25 to 50 years without breaking
- Easy to repair when it breaks, which implies that there are spare parts and
the design is simple enough to be understood by most people
- Long battery duration (12 hours at full cpu usage?)
- Be light and small, so that it is easy to carry everywhere
- Able to plug devices into it (keyboard, screen, ...)
- Already equipped with a small screen and keyboard (which might be touch screen)
to be usable everywhere
Goals and use cases
Now for the complex part. Let's list all my use cases. *insert blob sweat emoji*
- large data storage
- game programming
- 3d gaming
- 2d graphical gaming
- terminal gaming
- tabletop gaming helper
- emails
- gemini browsing
- web browsing (usually searching for information without knowing which website
will give me good results)
- reading
- writing blog and text files
- todo list
- watch videos, with subtitles
- spaced repetition flash cards
- view images
- draw images
- edit videos
- listen to music
- synthetise music
- arrange music
- compose music
- develop AI
- do/receive phone calls
- video conference
- send/receive text messages (either through cellular network or wifi)
- use AI program to split songs into tracks
- use VR for creative work and gaming
- calendar with automatic event repeats
- CI controller and agents
- system virtualisation (qemu)
- keyboard inputs methods (japanese)
- calculator
- 3d modelling
- password manager (though it would be nice to only use ssh keys instead of
passwords...)
- screen recorder
- run software that interacts with hardware where drivers only work on windows
- run software concurrently, for example with long running tasks
- route audio from program to program in a somewhat arbitrary way
- stores music or can access music on another system
- connect/manage other machines through ssh
- connect/manage other machines through windows remote desktop (rdesktop)
- recipe storage and querying
- inventory manager
- use AI code generation
- use git
- use github (create pull requests, check issues, etc)
- online banking
Wew, that was a lot.
It's quite clear that we can't have a forever computer that does all this
currently: it would be huge and expensive!
Let's assume that all use cases are important and cannot be deleted, and see
how different things can/can't deal with them.
Y = it can
N = it cannot
The Paper Sheet
- N large data storage
- N game programming
- N 3d gaming
- N 2d graphical gaming
- N terminal gaming
- Y tabletop gaming helper
- N emails
- N gemini browsing
- N web browsing (usually searching for information without knowing which
website will give me good results)
- Y reading
- Y writing blog and text files
- Y todo list
- N watch videos, with subtitles
- Y spaced repetition flash cards
- Y view images
- Y draw images
- N edit videos
- N listen to music
- N synthetise music
- N arrange music
- Y compose music
- N develop AI
- N do/receive phone calls
- N video conference
- N send/receive text messages (either through cellular network or wifi)
- N use AI program to split songs into tracks
- N use VR for creative work and gaming
- Y calendar with automatic event repeats
- N CI controller and agents
- N system virtualisation (qemu)
- N keyboard inputs methods (japanese)
- N calculator
- N 3d modelling
- Y password manager (though it would be nice to only use ssh keys instead of
passwords...)
- N screen recorder
- N run software that interacts with hardware where drivers only work on windows
- N run software concurrently, for example with long running tasks
- N route audio from program to program in a somewhat arbitrary way
- N stores music or can access music on another system
- N connect/manage other machines through ssh
- N connect/manage other machines through windows remote desktop (rdesktop)
- Y recipe storage and querying
- Y inventory manager
- N use AI code generation
- N use git
- N use github (create pull requests, check issues, etc)
- N online banking
12 Y
35 N
So paper can cover 25% of our use cases. Not too bad!
Terminal Only Kiss Linux
I have some experience using a terminal-only environment in 2022.
Let's see how our many use cases that works for.
- Y large data storage
- Y/N game programming (depends what kind of games!)
- N 3d gaming
- N 2d graphical gaming
- Y terminal gaming
- Y tabletop gaming helper
- Y emails
- Y gemini browsing
- Y/N web browsing (very hit or miss, lots of website break)
- Y reading
- Y writing blog and text files
- Y todo list
- Y watch videos, with subtitles
- Y spaced repetition flash cards
- Y view images
- N draw images (except pixel art)
- N edit videos
- Y listen to music
- Y synthetise music (but probably requires adapting software)
- Y arrange music (but requires creating a terminal DAW)
- Y compose music (but requires creating software)
- N develop AI
- N do/receive phone calls (might be possible but I don't know how)
- N video conference
- Y send/receive text messages (either through cellular network or wifi)
- Y use AI program to split songs into tracks
- N use VR for creative work and gaming
- Y calendar with automatic event repeats
- Y CI controller and agents
- N system virtualisation (qemu)
- Y keyboard inputs methods (but don't know how)
- Y calculator
- N 3d modelling
- Y password manager (though it would be nice to only use ssh keys instead of
passwords...)
- Y screen recorder
- N run software that interacts with hardware where drivers only work on windows
- Y run software concurrently, for example with long running tasks
- Y route audio from program to program in a somewhat arbitrary way (but might
be missing a jack terminal frontend to do it)
- Y stores music or can access music on another system
- Y connect/manage other machines through ssh
- N connect/manage other machines through windows remote desktop (rdesktop)
- Y recipe storage and querying
- Y inventory manager
- Y use AI code generation (but not sure how, I don't use it yet)
- Y use git
- Y use github (there's cli tools for it)
- N online banking
34 Y
15 N
69% (nice) of use cases. Considering this is only using a terminal interface,
it's impressive!
If we add the ability to have a 2d interface, window manager and pointing
device but no graphics card, 7 N becomes Y for a total of 83% use cases.
What to do?
We often say: "use the right tool for the job."
Which we could also rewrite as "use the simplest tool that does the job just
as well as other tools".
But there's an issue with that: we lose convenience.
Ever did construction and you need to go back and forth multiple times to grab
different tools you forgot?
Using the simplest tool for the job would mean walking around with a pen,
notebook, simple computer and having a complex computer at home.
How inconvenient! There's a reason phones have an integrated camera, it's
to not walk around with 2 devices!
My understanding of the forever computer is that we want to make most of our
use cases portable: move them from our big laptops/big desktops into
a single portable and convenient device. And we want that to have a long
battery life and be future proof.
This isn't going towards simplicity, it is the same idea as what we saw in the
introduction: we are going forward, towards doing more and having more
convenience.
If you are ready to sacrifice use cases (as I did when I went full time on
kiss linux for a while), what will happen when you get bored?
The appeal of the new technologies is strong, because that's where all the
new shiny toys are at. That's where most of the fun (and frustration) is at.
And don't forget that we can't have it both ways. If we have both a forever
computer and a complex computer for the remaining use cases, it will be
extremely tempting to do everything on the complex computer. Why?
Because it is convenient to do everything on the same device.
Instead of starting to talk about hardware, I'll instead end this post on
more personal questions.
Are you ready to sacrifice use cases in the name of stability, security and
simplicity?
Are you ready to sacrifice convenience by using the simplest tool for the job?
Are you fine with being left behind? Or do you prefer the attraction of the
shiny new things?
How much effort are you willing to put into a system where you will gain
control and understanding, but lose on value (use cases solved by it)?
Let us know your thoughts in the mailing list
~lioploum/forevercomputer@list.sr.ht
or by making a post on your own blog
or by sending an email to jojolepro [at] jojolepro [dot] com
Thanks you for your time, I hope this was an interesting read!
(CC0) Joël Lupien 2020-2022
View page source