Re: Thought experiment on the experience of hardware repair

Damien Slater <damien@slater.email>
Message ID
DKIM signature
Download raw message
//sending again using plain text

I’ve thought about this for some time also, it must be something about the outbreak times, serendipity etc.

Perhaps like some of you, my notes are important to me, they’re like my second brain and that’s what started me on the forever-compute journey; a computer for hosting my notes that would last at least as long as me = ).

Some of my conclusions are:
- 1. the solution probably doesn’t need a screen. I have 2 or 3 screens with me at all times (phone, laptop, e-reader). I don’t need another screen just for notes.  I’m willing to bet there will still be screens of some kind for the next 50 years.  The solution can self-host via browser and/or ssh. Screens on hand can connect to this server locally via bluetooth,wifi,USB.  
- 2. the solution should be repairable (agree with your discussion). We can design a PCB to last a long time (10 - 20 years) but longer than that becomes increasingly challenging. 
- 3. source components locally. I think optimising for repairability over 50 years means building from parts which are easily built regionally. For most electronics (capacitors, resistors) this is reasonable, but not every region CPUs are made equally (no TMSC cpus = ) ).  I have been considering low powered arm CPUs and punchy-microcontrollers (esp32, pi-pico). My current solution is proot-linux image running on an old android phone. Proot-image can be easily backed up and transferred to any other android phone.  Do I think android will be around in 50 years? Hard to say, unix will though.

CPU discussion
FPGAs are interesting, though they are typically underpowered compared to a traditional CPU.  It would be interesting to see if we can get a software solution to fit onto a micro-controller level of performance, this would make an FPGA solution more of a reality. It's also reasonable to predict we could configure an FPGA in 50 years, to behave like an FPGA today. To fit into this small performance footprint, we might look at something like uxn from 100r.co . Those guys have a comparable goal.  You can get more powerful FPGAs but can they be sourced regionally? Are we giving up too much power efficiency for flexibility?

Forever Computer should be for solving things which require decades of stability; this isn’t every compute use-case.  Heavy processing should probably be on the fastest newest technology economically possible. Where Forever Computer makes sense are for things like personal knowledge,  email, notes, contact management, journal; I want these things with me all the time, forever.

Looking forward to future discussions and will add more in the weeks/months to come.
Happy holidays
Reply to thread Export thread (mbox)