💾 Archived View for zaibatsu.circumlunar.space › ~solderpunk › phlog › the-standard-salvaged-computi… captured on 2024-07-09 at 00:45:02.

View Raw

More Information

⬅️ Previous capture (2023-03-20)

-=-=-=-=-=-=-

The standard salvaged computing platform
----------------------------------------
In a previous post[1] I argued that "sustainable computing", for all
the buzz and excitement around the term, is more or less an oxymoron,
that genuine sustainability actually calls for an urgent halt to the
manufacture of new electronic devices, and that the real long-term
future of computing consists of figuring out how to make the best
possible use we can out of the literal millions of devices which
already exist.

I don't expect this to be a popular opinion, but I do think it's
right.	People talk about "sustainability" as a Boolean concept -
something is either sustainable or unsustainable.  This only really
makes sense if there is a specific timeline attached to the concept,
and this is very rarely made explicit.	If we are talking about
sustainability on the scale of millions of years - and this is by no
means unreasonable, there are plenty of extant species today which
have lived in balance with their environment unchanged for tens
or hundreds of millions of years - then solar power and electric
cars and recycled plastic and LED lightbulbs and anything any of us
would recognise as "a computer" all fall pretty squarely into the
"unsustainable" bracket.  The mainstream sustainability movement
is actually not very ambitious at all.	That doesn't necessarily
undermine it.  The stark truth of the matter is that many aspects
of the modern lifestyle are "sustainable" on a timeline of mere
decades or perhaps a few centuries.  It's hard to consider that
anything other than a genuine emergency.  As a tactical, rather
than strategic, response to this emergency, "greening up" the
underlying technology by a factor of 10 such that the lifestyle
becomes sustainable on a timeline of centuries or perhaps even
a millennia makes a hell of a lot of sense.  But if we genuinely
care about "deep sustainability", and about the truly long term,
we can't lose sight of the fact that these changes are nothing
but temporary stalling tactics.  Planning *too* hard for the
next phase is arguably unwise, as the future is famously hard to
accurately predict.  But it seems silly to give it no thought.
Tactical decisions should be informed by clear strategy.  Hence,
I'm keen to hash these ideas out somewhat fully, starting with the
thought experiment of what happens if all the semiconductor fabs shut
down tomorrow and all we have forever after is what we have today.

Naturally, the salvaged computing landscape is going to be pretty
heterogeneous as we make do with whatever we can find.	But there
will probably be a clearly dominant paradigm, just like there has
been for the entire history of personal computing.  As much as it
pains me to admit it, if the shutdown happens anytime soon then the
most likely candidate for a "standard salvaged computing platform"
is the smartphone or tablet.  Despite all their shortcomings as
"real computers", these devices have a number of extremely important
advantages in any kind of hypothetical scenario where the trappings
of unsustainable industrialism are gone.

One, of course, is simple ubiquity.  The vast majority of households
in developed nations today have at least one of these devices and
probably several.  They are produced in such quantities, and the
upgrade cycle so short, that many houses will have old and unused
phones or tablets sitting in a drawer.  They outnumber desktop and
laptop computers substantially.

Another is that by design they have their display mechanism and
their input mechanism integrated into the same physical package.
Desktop computers, videogame consoles and SBCs like the Raspberry
Pi are all of comparatively limited utility without a TV screen or
a monitor (power hungry things!) and without a keyboard or a mouse.

The real killer, though, of course comes down to power.  It's
not just that phones and tablets have relatively modest power
demands compared to all the other contenders, it's the remarkable
uniformity in the power interface.  Every single vaguely recent
phone or tablet is happy to receive 5V 1A DC power over a standard
microUSB connection.  Oh, no, wait, that beautiful statement *used*
to be true, but we can't have a good thing for ever so now it's
"over a standard microUSB connection or this execrable new USB
C thing".  Compare this situation - one voltage, two connectors -
to the laptop landscape, where not only are power demands higher,
but just about every manufacturer has their own proprietary connector
(or several), and even different models by the same manufacturer
require slightly different voltages.  If we are to move away from
a centralised AC power grid (which in most parts of the world is
powered primarily by burning fossil fuels), and toward a world
where power is generated in a "just enough, just in time" fashion,
very close to the point of consumption - using pre-existing solar
panels, improvised wind generators using electric motors salvaged
from cars or household appliances, bicycle dynamos, etc., etc., then
having highly standardised power requirements is a tremendous help.

None of this is to say the other options have no merit and won't
find use, but it seems pretty clear to me that the phone/tablet
platform represents the path of least resistance in this vaguely
sketched out scenario.

There are, of course, downsides to this platform, most obviously that
it is designed with the intention of Apple or Google acting as middle
men between users and software developers.  Android devices fare
better here, in that even without rooting the device it is possible
to "sideload" APKs without a functioning internet connection and
without Google's involvement or approval.  But in general, working
toward getting as many of these machines running on "proper",
community maintainable, unrestricted operating systems can only be
considered vitally important work for the vision of salvage-based,
rather than manufacturing-based, computing.  The PostmarketOS
project[2] - which works toward putting a real Linux distribution,
not just Android, on mobile devices, to give them a useful life after
their manufacturer stops providing software updates to strong arm
people into buying a newer device - is the best example I'm aware of
of the kind of work that needs to be done.

Of course, if we step away from radical thought experiments and
back toward plausible very near future reality, where grid power
is still a thing and "real computers" are still perfectly viable,
a similar case can be made: operating systems which continue to
be bootable and usable even on "old" computers - especially 32bit
computers - and software which runs on those operating systems,
should be considered very valuable things to develop and to maintain.
I've already written about how I think the internet as we know
it is fundamentally unsustainable on any kind of long timeline,
but it's probably going to be around for the foreseeable future,
so ways of using the internet which still work perfectly well on
10 or 15 year old 32bit machines should be cherished and promoted,
and stuff which doesn't should be disparaged.  That means championing
things like Gemini and Gopher but not the web, things like email and
IRC and XMPP but not Slack, etc.  We should also be thinking about
different, better ways to use the internet, and even alternatives to
the internet itself, which are more oriented toward intermittent,
slow network connections, and people are certainly doing this (see
Secure Scuttlebut, see the dat:// protocol, see IPFS, and probably
many more things that I'm not even aware of), but we should also
be writing software implementing these new technologies which runs
comfortably on machines which are 10 years old or older, which
I don't necessarily think people *are* doing.  A lot of software
in the Scuttlebut ecosystem, for example, is written in Electron,
a famously resource guzzling framework.  If the user interfaces
for these new tools are more demanding to run than the underlying
cryptography, that's a big problem.

At this point, I feel like a case can be made for the following
practices as good "solarpunk praxis" (with an awareness that I'm
stretching the term "solarpunk" a bit, as it usually denotes a
degree of technological optimism I'm not sure I share - but it
seems to be the magic term to use to get noticed by the people I
want to have dialogues with):


  absolutely "have to".  Buy used instead.  Only decreasing demand
  will slow down manufacturing, and only slowing down manufacturing
  will substantially reduce environmental impact.  This is the single
  most important point in this list.  It extends even to things
  "normal people" wouldn't consider computers - fitness trackers,
  voice activated personal assistants, smart lightbulbs, etc.

  and/or charges directly from 5V DC (or perhaps 12V DC): they will
  be more useful in a world without ubiquitous AC grid power, which
  is what true sustainability looks like.  This is more important
  when buying new - capitalist manufacturers will naturally shift
  toward manufacturing more of what sells and less of what doesn't.
  Exercise this very real power for good - but only as a last resort.

  Android devices over iOS devices: they will be more useful in a
  world without Apple, Google or ubiquitous internet, which is what
  true sustainability looks like.  This is more important when buying
  new, as above.

  user serviceable over those which aren't.  The manufacturers are
  almost never your friends in this, so this is mostly a matter of
  choosing the least bad option. This is more important when buying
  new, as above.

  old unused devices which still run but are just "old" or "slow" or
  are unsupported.  The inefficient and environmentally unfriendly
  electronic recycling process will turn a working computer into a
  non-working computer, which is never a winning proposition.
  Instead, give them new life by installing PostmarketOS (on mobile
  devices) or a specialist OS targeting old hardware (on non-mobile
  devices).  If you do not have the skills or interest to do this,
  try to donate the device to somebody who does.  Places like
  hackerspaces should play an active role in this - teaching skills,
  collecting devices, reflashing devices, and distributing devices.

  or unsupported computing devices, or which try to "jailbreak" or
  otherwise liberate locked down devices to make them more general
  purpose, and if you can contribute to them by donating hardware,
  writing code, writing documentation or donating money, do so, to
  whatever extent your skills and circumstances allow it.  These
  projects are far more important than projects to design and
  manufacture new, "better" hardware in the interests of
  sustainability.  Manufacturing is the *problem*, it cannot be a
  large part of the solution.

  which is 10 years old or older (if you don't already have one),
  and see how well your software runs on it, if it runs at all.
  If the results aren't impressive, try your best to improve them.
  Learn lessons from the experience, apply them to everything you
  write going forward.  If your favourite operating system doesn't
  run on the machine at all, find a new favourite out of the ones
  which do, and use it even on your newer machines.  Push back hard
  against the tendency for software "progress" to deprecate working
  hardware.

  software for Android, even if in an ideal world you'd really rather
  not.  It seems likely to be by far the most common usable computing
  platform in an off-grid world.  Distribute your software only
  through FDroid or by directly distributing signed .apk files
  yourself.

  without relying on the Google Play Store, learn how.  Teach your
  friends.

  and efficient off-grid 5V DC supplies in as many different ways as
  you can.  Share your results widely.  Develop instructions which
  can be followed by people with minimal skills and tools.

[1] gopher://zaibatsu.circumlunar.space:70/0/~solderpunk/phlog/discussions-toward-radically-sustainable-computing.txt
[2] https://postmarketos.org/