More on withered technology, part I
-----------------------------------

Back in April I wrote about Gunpei Yokoi's philosophy of "Lateral
Thinking with Withered Technology"[1], with prompted Anna (agk)[2]
and The Free Thinker (freet)[3] to write responses of their own.
I have been meaning to write them both responses ever since but got
distracted writing about other things first.  It's taken me much
longer than I wanted to get responses out, and I'm still struggling,
so I'm switching to an instalment plan strategy rather than trying
to find the time to get one giant response that covers everything
into a state I'm happy with.  This post will just be my response to
freet's post.  It's not so much a detailed point-by-point response
as it is my long musings on aspects of this whole permacomputing
thing which have arisen as a consequence of my reading their posts,
and that'll be true for my response to Anna (hopefully next weekend)
as well.

Of the two responses, freet's was the one with the most with
concrete instances of using various particular withered computing
devices in novel ways.	Specifically, video game consoles (a PS2
and the Nintendo DS) were mentioned, as were routers and other home
networking devices which can (or could) run OpenWRT.  I think both of
these categories of withered device warrant serious thinking about.
They have historically been kind of off my radar but now are very
much on it.

All the way back in 2020, I wrote a post about the so-called
"standard salvage computing platform"[4], arguing that we ought to
think of Android smartphones as constituting this.  My reasoning was
simply that these were absolutely ubiquitous, approximately free
(most people, if they really tried, could probably score an old
and somewhat-out-of-shape but functional smartphone off a friend of
family member who has it sitting in a drawer somewhere, for free or
in exchange to buying them a drink or whatever), had integrated I/O
(the touchscreen being built right into the device serving as the
primary channel for both), had lower power requirements and also
very well standardised physical connections for power.

All of these advantages remain true, but in the intervening years I
have come to think I probably down-played some obvious downsides
as well.  Probably the biggest of these is radical hardware
variability.  There's some sense in which "Android smartphones"
is a meaningful device category to talk about, but it's also
undeniably true that at the level of chipsets and bootloaders, we
are actually talking about easily hundreds and probably thousands
of distinct and incompatible devices.  These devices are able to
run a Linux-derived operating system at time of launch only due to
extensive phone-specific porting efforts carried out by engineers
employed by the phone's manufacturers.	Those engineers have full
access to all the details of how the phone are built and they don't
have to worry at all about bootloader security or anything like that.
Volunteer communities trying to revitalise those devices after they
have abandoned by their manufacturers are working in the dark and
against obstacles.  The effort involved in porting later versions
of Android, or alternative operating systems like PostmarketOS,
to those devices is substantial.  The workload is multiplied by
the total number of distinct hardware designs in circulation.
The total is, realistically, impossible for a niche movement.

The video game consoles landscape looks radically different to this.
These devices are manufactured at very large scale (the PS2, Nintendo
Switch and Nintendo  DS each sold over 150 million units worldwide)
but the hardware variation, while non-zero (you don't just have
things like the "fat" vs thin PS2, but there are a few different
generations of fat PS2 hardware, as I found out recently I when I
naively thought I could transplant an optical drive from one to the
other), is negligible.	All 150 million of the devices branded with
the same name are 100% binary compatible - customers would riot if
the expensive discs and cartridges they bought stopped working on
later generations of the same hardware or after a firmware update
was installed.	Figuring out how to repurpose the five most popular
consoles of this century is guaranteed to net you many, many more
repurposed devices than targetting the five most popular makes and
models of Anrdoid phone would, no question.

Not only is the amount of reverse engineering work required to
revitalise all those devices less than that required to revitalise
all the smartphones, the incentives for doing so are very different.
People jailbreak phones for idealistic reasons around freedom and
privacy and environmentalism (as well, of course for the fun of the
technical challenge).  People do the same to video game consoles
to save money or make money by pirating games, which is something
that appeals to many orders of magnitude more people than casting
off corporate shackles does.  When many more people are working on
far fewer devices, the work naturally gets done faster.  In fact, an
awful lot of the work is already done!	Many of these platforms have
had their security systems broken and can now run arbitrary code,
including non-game homebrew software.  Did you know you can even
install the Libreboot Open BIOS on the original Sony Playstation?
It's not necessarily easy, it's a hardware hack[5], but still!

Note only are there homebrew development kits out there for a lot
of these platforms, the aftermarket hardware care and support is
surprisingly well-developed, too.  There are a lot of videogamers out
there who like old machines but are not themselves hardware hackers,
and this makes it commercially viable for folks who *are* hardware
savvy to operate online stores selling refurbished devices which have
had e.g. new batteries installed in them.  There are people making
and selling replacement plastic shells for all the handheld devices.
You can buy kits to upgrade the display in original GameBoys to be
brighter and clearer!  I was really surprised when I discovered
all this earlier this year.  This is real permacomputing praxis,
breathing new life into old machines, including machines from last
century, and it's being done for entirely non-ideological reasons,
by people who just wanna play old games and enjoy them on original
hardware.

If it's not already clear, I have become *very* interested in
old game devices as salvage computing platforms.  The real turning
point was being shocked at just how low the barriers to entry are in
some cases.  Specifically, I bought a Nintendo DSi (a device which
has a (stylus) touch screen, WiFi, front and rear-facing cameras,
a microphone, an SD card reader and a Real Time Clock, so not one
million miles away from a rudimentary smartphone) much earlier
this year (before making any of these withered tech posts) after I
learned there there's an open-source C library[6] which easily exposes
just about all the important functionality.  You can compile your
homebrew DS software on a Linux machine using GCC and make, it's the
most conventional development experience imaginable.  You do need a
so-called "flashcart" to be able to get the resulting binary running
on hardware, but still.  The contrast with developing for Android
is tremendous, with the consequence that I've actually *done* some
DS development using libnds, whereas every time I've even thought
about developing for Android I've noped out after doing a little
research and not wanting to go through the hassle of even getting
set up for it.	I'll be writing more about DS homebrew in the future!

It's not all necessarily totally smooth sailing, of course.
The DS and DS-Lite support WiFi but not WPA2 networks.	There's an
off-the-shelf web browser for the DS but it can't do modern HTTPS
due to expired CA certificates (there's an open source browser
which can[7], but it's extremely rough and ready).  These are hardly
insurmountable obstacles, though.  A headless RPi or even an ESP32
device can provide a very low-power short-range WEP or open WiFi
network, possibly locked to MAC addresses, and can use HTTP to serve
crusty old CSS-renderings of RSS feeds, Gopher and Gemini content,
for example.  Old smartphones and tablets, naturally, could take
advantage of such a facility as well, this being much safer than
connecting them directly to the public internet.

Of course, RPis and ESP32s are not exactly in the "withered
tech" category, and indeed I'm encouraging permacomputing folk
to explore lateral thinking with withered technology instead of
building new widgets based around these dirt cheap and abundant
modern devices precisely because all these dirt cheap DIY-friendly
modern devices are powerful vectors for the "hardware grows on
trees and lil' computers are so cheap you can totally buy some
just for fun!" mentality which permacomputing needs to actually
oppose if it's to have any kind of coherence.  There is surely some
"breaking eggs to make omelettes" rationality in using such devices
if you already have them lying around, or can buy them cheaply used
from consumerist makers who have upgraded to a later version, and
apply them to breathing new life into yet older things which would
otherwise be hard to utilise, but the philosophical tension is there.
This bring us nicely to the other category freet mentioned, which
is consumer networking tech.

Things like modems, routers, NASes etc feel to me like they are
probably situated somewhere between game consoles and smart phones in
terms of how much hardware variability there is.  Like smartphones,
many of them are sold running something Linux-derived under the
hood, and as such it is unsurprising that many of these devices
are or were supported by OpenWRT.  I am far from an expert here,
but I *think* it's probably true that the proportion of popular
devices which are or were well-supported by OpenWRT is greater than
the proportion of popular smartphones which are well-supported by
PostmarketOS or LineageOS or anything like that.  Some very quick
research on this front revealed to me that a popular Linksys NAS
device, the NSLU2[8], manufactured between 2004 and 2008, was able to
run mainline or close-to-mainline Linux and *BSD, and seems to at
one point have had a pretty active and dedicated community of folks
doing such, affectionately referring to the device as the "Slug"
and sometimes even bestowing memory upgrades upon it.  Some really
quick poking around on eBay and my local Craigslist equivalent
suggests these Slugs are still not hard to find and dirt cheap.
There are probably plenty of other devices like this, too.  Many of
them probably have wireless hardware in them, meaning they could
also act as gateways for things like DSes, getting them easily into
Gopher- or Geminispace, bridging them to alternative networks like
Yggdrasil, etc.

Some of you might be wondering quite what the point of all this
is, when there are also still countless millions of "real" computer
devices out there waiting to be reused for even less effort.  This is
a fair question (one for which "fun" is a perfectly permissible
answer, of course), and one I will try to say something about when I
write my response to Anna, whose post got me thinking more about big
picture "why?"/"what for?" type questions than specific categories
of device type.

[1] gopher://zaibatsu.circumlunar.space:70/0/~solderpunk/phlog/lateral-thinking-with-withered-technology.txt
[2] gopher://sdf.org:70/0/users/agk/2026-04-14-lat.txt
[3] gopher://aussies.space:70/0/~freet/phlog/2026-04-17RE_Lateral_Thinking_with_Withered_Technology.txt
[4] gopher://zaibatsu.circumlunar.space:70/0/~solderpunk/phlog/the-standard-salvaged-computing-platform.txt
[5] https://libreboot.org/docs/install/playstation.html
[6] https://blocksds.skylyrac.net/libnds/index.html
[7] https://codeberg.org/SkyLyrac/bunjalloo
[8] https://en.wikipedia.org/wiki/NSLU2