pages tagged 0GFSFLAhttp://www.fsfla.org/ikiwiki/tag/0G.htmlFSFLAikiwiki2021-08-03T00:39:08Z0G captain's log 20190811http://www.fsfla.org/ikiwiki/blogs/lxo/2019-08-11-0G-log.en.html2019-08-18T16:05:32Z2019-08-11T22:45:55Z
<p>Captain's <span class="selflink">log</span>,
stardate <a href="http://www.fsfla.org/ikiwiki/blogs/lxo/2019-08-02-0G-log.en.html"><-</a> 1565561012.767 <a href="https://www.fsfla.org/~lxoliva/#0G">-></a></p>
<p>The 0G launch in pt_BR last weekend was quite well received, much
interest expressed throughout the rest of the conference. It's quite
nice when even the session photographer and a hotel waiter express
interest in the project <img src="http://www.fsfla.org/ikiwiki/smileys/smile4.png" alt=";-)" /></p>
<p>The night before the conference, the notion came up of a long-range
radio, operating at 27MHz for low-bandwidth, high-power and very
long-range (up to 500km, good for a sparse community, for getting
stranded on the road and reaching your other device back home to ask
for help, etc) communication, possibly negotiating alternate
frequencies for faster but shorter-range communication. It feels
pretty much essential to me now, but it's likely to take some time for
our (GNU) radio expert to develop, so a Caninos Loucos first batch is
absolutely unlikely to have them. Prof Zufo tells me LoRaWAN is
something they have in one of their options of boards, but I'm told
that's proprietary to the core.</p>
<p>Some concerns about the power use of such low-frequency radios were
raised; if we were aiming at truly embedded processors and operating
systems, such as FreeRTOS or eCos on AVR, to get weeks or months of
battery life, they'd be more worrying, but my goal is to run GNU,
which likely means a full-fledged MMU-capable CPU, so I guess
less-than-weeks of battery life will have to be tolerated <img src="http://www.fsfla.org/ikiwiki/smileys/smile4.png" alt=";-)" /></p>
<p>There seem to be higher frequencies than 27MHz available for our kind
of use world-wide, but given the shorter ranges that seldom span
different jurisdictions, the exact available frequencies might vary.</p>
<p>Designing radios that could work at multiple such frequencies,
especially ones that could always listen at 27MHz for others' hails,
appears to be quite a challenge. Maybe it would make sense to have
separate radios to operate at different frequencies, but that could
get expensive. Wing development might involve a wien bridge for such
a wide range of frequencies... Plenty of conversation about that
going way over my head; mentions of hackrf and rad1o.</p>
<p>Still no response from Puri.sm, but I got an email from someone (at or
close to?) Pine64 about running 0G on Pinephones. The rtl8723bs WiFi
might make it a challenge to realize the vision of a blob-free system,
but it would be nice to bootstrap/grow the network with such devices,
and getting a modern phone-oriented GNU system to adapt to 0G would
certainly be welcome. Plus, if this WiFi chipset is widely deployed
and has an expected lifetime longer than usual, it might make sense to
undertake the effort of at least trying to develop Free firmware for
it.</p>
<p>So blong...</p>
0G captain's log 20190802http://www.fsfla.org/ikiwiki/blogs/lxo/2019-08-02-0G-log.en.html2019-08-18T16:05:32Z2019-08-03T02:24:10Z
<p>Captain's <span class="selflink">log</span>,
stardate <a href="http://www.fsfla.org/ikiwiki/blogs/lxo/2019-07-26-0G-log.en.html"><-</a> 1564800237.752 <a href="http://www.fsfla.org/ikiwiki/blogs/lxo/2019-08-11-0G-log.en.html">-></a></p>
<p>I asked for leads/contacts at businesses that might be interested in
collaborating, particularly on the hardware part, say Purism and
Pine64. lkcl was named for EOMA68 (no response from him yet; I wonder
if he even received my email about it...); other mentions were GnuBee
(lovely name) NAS, RaptorCS (POWER for handsets?!? yeah, probably
not).</p>
<p>One interesting thought was to separate the requirements for
(development and) validation from the actual goal. Like, even a
freedombox without screen or WiFi could work for validation of the
basic networking infrastructure. An SBC plus a touchscreen, and WiFi
(builtin or dongle) would get us closer to the target environment.</p>
<p>Some conversation about CPUs concludes that x86 is out (ME, PSP),
POWER is too power-intensive, RISC-V is coming but the only PC with it
is too costly, and ARM performance is on par with that of our old and
beloved Librebooted Thinkpads. But we're not aiming at a PC, but at a
mobile, ultraportable communicator. So ARM surely can do, and RISC-V
is a decent bet for the not-so-distant future.</p>
<p>GNUnet was brought up again, and we wondered if it offers
de-localization like Tor. Having two separate calling addresses and
separate infrastructures might be a plus or very very confusing. But
it seems that GNUnet already has anonimized calls, whereas getting GNU
jami to work behind Tor would require significant effort. Would
GNUnet work behind Tor?, I wondered. Would that even make sense?
Nobody in the channel had an answer handy. Asking on the gnunet
channel is added to the todo list. Gotta get in touch with them
anyway. It's looking like a promising avenue to explore for 0G.</p>
<p>Conversation returned to readily available boards. Orange Pi (not
using its blob-ridden WiFi) might work for validation, but for the
real thing we'd want something more like EOMA68's LibreTea, or an
OLinuXino (that's what's in Olimex's freedombox, and Olimex has
touchscreens and WiFi for sale as well. Gotta check that their WiFi
dongles are freedom-respecting, and whether the touchscreens could be
connected to the freedombox, or just to the barebones boards). Golden
Delicious, if still active, might help with the design of custom
boards.</p>
<p>Mentioning Freedombox in the lecture got us a mention at
https://freedomboxfoundation.org/news/debconf_2019/</p>
<p>Adding a touchscreen to a freedombox would get us something pretty
close to a minimalist 0G device (as in, no camera, no speaker or
microphones, no GPS, no WiFi), but possibly something to use for
development, testing, and even to enable users to enter a
full-disk-encryption passphrase. Come to think of it, any reason not
to run a freedombox on your own 0G uPoC? Rather than having it only
at home, have one in your pocket as well! Surely you'd rather have
full-disk encryption for the latter, and I'm pretty sure you'd then
have a touchscreen to enter the passphrase too!</p>
<p>(Alternate way for networked machines to get boot-up passphrases: (a)
initrd https server for a client to supply the passphrase, (b) a
<a href="https://madresistor.com/diy-vt100/">free hardware serial interface</a>,
(c) a pluggable USB keyboard (if you know it's prompting for one;
or... keyboards with screens anyone?), or (d) extend DHCP to supply
the boot-time passphrases, and extend GRUB to ask for them before
prompting the user. Then as long as any of the DHCP servers on the
network remains up, the rest of the network can be brought back up
without manual intervention. But if all of them are brought down and
taken elsewhere, someone would have to enter a passphrase to bring any
of them back up. Something to implement one of these days, when my
todo list is empty... <img src="http://www.fsfla.org/ikiwiki/smileys/smile.png" alt=":-)" /></p>
<p>I saw links about
<a href="https://www.crowdsupply.com/libre-risc-v/m-class">related</a>
<a href="https://www.vice.com/en_us/article/paj8z8/a-diy-internet-network-has-drastically-expanded-its-coverage-in-nyc">news</a>
elsewhere that I shared with the channel.</p>
<p>A conversation on the short range of WiFi ensued, and a pet project
named Wing was disclosed: a libre 900MHz radio modem for GNUnet. We
immediately adjusted the acronym to Wing Is Nil G. Though I can't
really imagine something without WiFi would fly, the addition of
longer-range Wings could enable uPoCs to get connectivity in less
dense areas. Why not have both?</p>
<p>A question I answered in the <a href="https://debconf19.debconf.org/talks/72-whos-afraid-of-spectre-and-meltdown/">DebConf19
Spectre&Meltdown</a>
speech was brought up: I mentioned there something to the effect that
I didn't know enough about CPUs to promise that any of the Freer ones
we hope to use didn't have software blobs hiding deep in the hardware,
playing the "indistinguishable from a hardware circuit" role, but it
was pointed out that maybe the questioner was looking for an answer
along the lines of "indistinguishable from hardware", even though the
topic of the speech was security, and with that framing,
indistinguishable from hardware hardly cuts it given that even
hardware may exhibit malicious features and vulnerabilities. Anyhow,
among ARM and RISC-V I should probably have mentioned OpenPOWER.</p>
<p>Speaking of RISC-V, a rumor was mentioned that some (early?) vendors
had licensed DRAM controllers that required non-Free init blobs. None
of the community-oriented designs had taken that path, but it's
something to be watchful for to avoid later headaches. Apparently
Sifive's PC <a href="https://forums.sifive.com/t/ddr-controller-configuration-register-values-for-hifive-unleashed/1334">had it (now
fixed)</a>.
Phew!</p>
<p>Time is ripe is a phrase I have often written and spoken WRT 0G, and
that's what I felt when I read <a href="https://puri.sm/posts/what-a-no-carrier-phone-could-look-like/">this post by the Purism
CEO</a>,
about getting phone service over WiFi or a cell data link. Looks like
jmp.chat is getting a competitor in Librem Dial, which may mean 0G is
getting another alternative for phone service. The post had the GPG
fingerprint for CEO Todd Weaver's keys, so I fetched them and sent him
email, at last. Let's see whether they're interested in some form of
cooperation, or even in just building on the 0G ideas or software that
could make Librem 5 an even more privacy-oriented device. I'm looking
forward to his answer.</p>
<p>The pt_BR launch of the 0G project is scheduled to 16:25 UTC-3 on
20190803, in São Paulo, at <a href="https://cfp.linuxdev-br.net/2019/talk/9NSA8J/">Linux Dev Conf
Br</a>. I've translated
the slides and improved them a little, so as to not forget to make fun
of idIoTic devices again.</p>
<p>Caninos Loucos folks are going to be at the conference this weekend,
I'm told, so I'm getting a chance to ask them F2F about the choices
for WiFi, and about the possibility of adding Wing to their upcoming
run of OpenMoko Freerunner-like devices, that we might be able to
piggyback on for 0G uPoCs.</p>
<p>Many interesting thoughts and developments over the past week, despite
my having been brought down by bug that was making the rounds at
DebConf19, and taken the entire family down with it. Keep them
coming! No, the ideas for 0G, not the bugs <img src="http://www.fsfla.org/ikiwiki/smileys/smile4.png" alt=";-)" /></p>
<p>So blong...</p>
0G captain's log 20190726http://www.fsfla.org/ikiwiki/blogs/lxo/2019-07-26-0G-log.en.html2019-08-18T16:05:37Z2019-07-26T20:47:24Z
<p>Captain's <span class="selflink">log</span>,
stardate <a href="http://www.fsfla.org/ikiwiki/blogs/lxo/pub/0G.en.html"><-</a> 1564110000.042 <a href="http://www.fsfla.org/ikiwiki/blogs/lxo/2019-08-02-0G-log.en.html">-></a></p>
<p><a href="http://www.fsfla.org/ikiwiki/blogs/lxo/pub/0G.en.html">0G</a> successfully launched from the surveillance blackhole on
Tuesday at <a href="https://debconf19.debconf.org">DebConf19</a>, and
interesting suggestions and conversations took place in the #0G IRC
channel on FreeNode. I'm taking note at least until we settle on a
home in cyberspace and set up an archived mailing list.</p>
<p><a href="https://wom.community/">WOM</a> was mentioned as a project with similar
goals. A <a href="https://livestream.com/internetsociety/radnets17/videos/164804778">recording of a
speech</a>
can be downloaded with youtube-dl. An
<a href="https://www.vice.com/en_us/article/595zg5/sopranica-jmp-wom-cell-network-diy-anonymous">interview</a>
about JMP, WOM and other Soprani.ca projects. Huge overlap indeed!</p>
<p>Using an 8-bit MCU could make for a device with battery lifetime
measured in weeks or months rather, than hours or days; it could make
and receive calls over WiFi or some other kind of radio, and have real
buttons and audio feedback, rather than a screen. It might be tricky
to enter Onion addresses on that, but if someone makes them, there's
no reason to rule them out.</p>
<p>Lower-frequency radios with longer ranges could be something nice to
build into our devices, in addition to or instead of WiFi, more so
considering the difficulties of regulation and of finding
FaiF-compatible WiFi these days. WiFi has the advantage of being
widespread, and being available in many existing handsets, whereas WOM
is peer-to-peer XMPP over chibiArduino (thin messaging on IEEE
802.15.4), operates at 900MHz, and reaches dozens of km with direct
line of sight. Still, being radios, both would require going through
radio licensing, even if using a publicly-available spectrum.</p>
<p>Other efforts to mesh WiFi were mentioned: the earlier <a href="http://www.servalproject.org/">Serval
Project</a> and some recent project by
Facebook whose name was not mentioned.</p>
<p>GNU jami (formerly GNU Ring) was mentioned for decentralized phoning.
It would take some changes to avoid disclosing a user's location/IP
address so as to not enable tracking, and to use Onion addresses to
initiate calls and whatnot (protocol changes to go from UDP to TCP
would likely be needed as well), but it's definitely in the radar.
Getting opendht to use TCP would be a first step.</p>
<p>Using <a href="https://www.ledsmagazine.com/smart-lighting-iot/article/16700551/london-connects-another-28000-street-lights-to-wireless-mesh-controls-updated">street light mesh
nodes</a>
was brought up.</p>
<p>Likely difficulties of sourcing ath9k_htc/carl9170 masks for use in
SoC designs were discussed.</p>
<p>Using <a href="https://gnunet.org">GNUnet</a> as an alternate network layer was
suggested. It can route and mesh over various physical layers
(radio-modems at various frequencies were mentioned), and there are
even <a href="https://gnunet.org/en/about.html">phone call</a> infrastructure
built into its higher layers! Wikipedia has somewhat more accessible
information about various aspects of GNUnet, including transport.</p>
<p><a href="https://caninosloucos.org/downloads/Caninos%20Loucos_datasheet_Labrador_R3_EN.pdf">Caninos
Loucos</a>,
a project by John Maddog Hal and prof Marcelo Zufo at University of
São Paulo, is likely to build OpenMoko Freerunner-like devices in the
very near future, maybe we could use them. GPU could be a problem:
they've reverse engineered the PowerVR and implemented new drivers,
but they're looking into whether they can publish them as Free
Software, according to an unscheduled (and not recorded) BoF at
DebConf19. WiFi/Bluetooth is probably an issue too.</p>
<p>JMP.chat is certainly interested in expanding to other countries, as
long as carriers support the required feature set (text messages,
picture messages, short codes and incoming calls) through a usable API
(REST, XMPP). So far only US and Canada are known to meet these
standards, but if we find about others, we should let them know.</p>
<p>This hopefully covers all highlights in the channel up to yesterday.
After some pondering, the IRC logs are not going to be published, so
please let me know if you'd like to be identified as the source of any
suggestion/information.</p>
<p>So blong...</p>
0G Manifesto: Escaping the Surveillance Blackhole with Free Mobile Computinghttp://www.fsfla.org/ikiwiki/blogs/lxo/pub/0G.en.html2021-08-03T00:39:08Z2019-07-23T01:43:29Z
<center>
<h1>
Escaping the Surveillance Blackhole<br />
with Free Mobile Computing<br />
0G Manifesto
</h1>
<h2>
Alexandre Oliva
</h2>
</center>
<p>Mobile phones have grown into nearly irresistible surveillance
devices. Very desirable features are bait that compels us to carry a
multitude of data-gathering hardware and software permanently
connected to a network that tracks us. The good news is that time is
ripe to split them apart, and build ultra portable communicators with
all of the desirable features and none of the surveillance. Here's
how.</p>
<h1>Introduction</h1>
<p>Modern "smart" phones are designed for surveillance: multiple cameras,
microphones, location tracking, permanent wireless networking, and
tons of data-collection apps disguised as useful programs turn them
into a blackhole that no personal data can escape.</p>
<p>They also offer such useful functionality that even privacy-aware
people find it very hard to resist them: looking up information on the
go, getting maps and directions, keeping in touch with friends, family
and business partners from anywhere are extremely valuable indeed.</p>
<p>All of these desirable features are currently available on laptops,
but their bulk, interaction modes and even power-up times can make
their use on the go not quite as convenient. How hard could it be to
build an instant-on, touchscreen "laptop" in a phone form factor, so
that we could carry it in a pocket rather than in a backpack?</p>
<p>That is part of the solution, but it's not the complete solution: that
will require some work on networking, software, and hardware. It's
all in-reach and doable today, and I refer to the combination of such
lightweight hardware, software and networking, that enable us to
escape the surveillance blackhole, as '''0G'''.</p>
<h1>Networking</h1>
<h2>Reachability</h2>
<p>You might have long ago concluded that, since you wish to receive
calls and messages, you have to be tracked anyway, at the very least
by the phone company. Rethink!</p>
<p>Tor, the Onion Router, offers Onion services, that have long been used
by e.g. The Pirate Bay to keep itself accessible without disclosing
its actual location. My own home server and my laptop can reach each
other no matter what network the laptop is connected with.</p>
<p>This is done in a way that does not require tracking or registration
with a centralized server: neither the Tor node chosen at random to
listen for incoming requests for any of these, nor the Tor nodes that
deliver the requests to them, can connect the service with the server
or its location. Let's use that to locate each other, then!</p>
<p>It may not be ideal for calls to always go through Tor, because of
latency and networking protocols, but once caller can reach callee and
they can negotiate a session, they may choose to communicate directly.</p>
<p>Software should be configurable to allow such policies as (a) direct
communication only with preauthorized and authenticated parties, (b)
all communication through Tor, and (c) redirect and even (d) reject
calls from unidentified callers, etc, besides end-to-end encryption
policies.</p>
<h2>Connectivity</h2>
<p>Just having a SIM card, or even a cellular link, is enough for base
stations to track you through your phone. Considering that commercial
cellular networks require compatible modems to run proprietary
software, that grant the network not just the ability to track the
user through the phone, but also to control the phone in various
undesirable ways, we're probably better off without them entirely.</p>
<p>We may have to work on increasing WiFi coverage, though. The One
Laptop Per Child project, for example, hoped to expand the reach of
school's networks through mesh networking in their famous XO laptops.
The mesh network technology they adopted required non-Free firmware
and was a source of no end of headaches to their engineers, so we'd
better steer away from that, but maybe ath9k_htc/carl9170 cards with
Free firmware could be built into 0G devices and reprogrammed to serve
this goal through ad-hoc (vs infrastructure) P2P communication.</p>
<p>Though 0G communicators are encouraged to not have cellular modems
built into them, nothing stops phones that do from joining 0G logical
networks, even while using a cellular data link.</p>
<p>Laptops and computers equipped with WiFi are also welcome to join 0G
logical networks, to share their local connectivity, and to enable
their users to interact with users of other 0G devices through calls,
messaging, etc.</p>
<h1>Software</h1>
<p>Existing phones with hardware switches to disable the cellular modem,
and running the GNU operating system, are very likely to be excellent
platforms for 0G development, testing, and use, and their software is
a likely basis for modifications towards 0G.</p>
<p>I don't envision 0G-specific apps, but rather modifications to
existing apps, e.g. to use Tor networking and addresses instead of
phone numbers, policies settings, network sharing, and MAC address
randomization.</p>
<p>User interfaces aimed at use of cellular networks should not be
removed, but extended, so that it remains functional on equipment that
has built-in or add-on cellular modems.</p>
<p>The possibility of directing calls to legacy phone numbers through
Internet-based call forwarding proxies should be present eventually;
<a href="https://jmp.chat">jmp</a> service is one way to gateway SMS and voice
through XMPP and SIP, and at least SMS over XMPP should avoid
surveillance over Tor; whether voice does is TBD.</p>
<p>Decentralized, serverless networking apps should certainly be given
higher visibility than centralized counterparts, as long as they do
not enable user tracking, and built-in support for use of a Freedombox
for backups and whatnot (over Tor) should be aimed for.</p>
<p>Apps that are little more than web sites plus local tracking are not
welcome. Indeed, we should highly encourage functional websites with
graceful degradation when Javascript is not granted permission to run
and consume network, memory and cpu. It is unlikely that surveillance
businesses will target our system any time soon anyway, and we might
even give them additional reasons to respect our freedom or leave us
alone, with some strong copyleft userland component.</p>
<h1>Hardware</h1>
<p>Although existing phones can be used as 0G ones, escaping the
surveillance blackhole and realizing the 0G networking vision are more
securely achieved with custom Free Hardware running exclusively Free
Software. That is not as economically prohibitive as it used to be,
and the growing availability of Freedom-respecting hardware further
invites to explore the design space and make further convenience
improvements and satisfy additional goals.</p>
<p>For example, fitting an EOMA68 board in a smart phone case with
touchscreen, Free WiFi, and battery would make for an interesting 0G
phone. More so if it's a next-gen board with a RISC-V CPU and a
libre, accelerated-3D GPU!</p>
<p>Now, if you're like me, and get home and plug the phone into a USB
port of the laptop to recharge and whatnot, what if the laptop was
just a case with keyboard, touchpad and screen, and the plugging in
connected the communicator not just with the battery and power supply,
but also with the larger screen (or screens), keyboard, touchpad,
wired network, USB ports, storage, speakers, microphones, webcams,
etc?</p>
<p>What if you could connect a (USB3?) video output to the communicator,
and then to a projector, to show the slides for your presentation
without having to carry even the laptop case around?</p>
<h1>We can do this!</h1>
<p>I envision coopetition between multiple suppliers in designing and
selling handsets, and widespread collaboration in developing the
software changes, towards realizing this vision.</p>
<p>One strategy I suggest, highly inspired in the OLPC model, is to build
reasonably inexpensive 0G handsets, in robust, cute and repairable
cases, like the XO, targeting it at school kids, and selling them in
large quantities for governments to offer them in public schools. As
in the OLPC plan, the school network gets extended throughout the
neighborhood by the meshed handsets, serving the entire community.
Governments would not want to subject kids to corporate surveillance,
so make them inexpensive enough that kids won't be at significant risk
of having their phones stolen from them, and avoid corporate
"partners" sabotage, and you should be able to get the network
bootstrapped. In communities with broad network coverage, there will
likely be more buyers of fancier, higher-margin 0G handsets.</p>
<p>After so many years of GNU development, we are very close to making it
usable in the most prevalent computing form factor so that it can
liberate people from the chains that existed when it started, and also
from the newer threats arising from surveillance capitalism.</p>
<p>Let's make it happen! Join channel #0G on Libera.Chat IRC, or get in
touch through my FSFLA email address, at least until the project finds
a home.</p>
<hr />
<p>Thanks to my friends quiliro, jxself, figosdev, cascardo and sergiodj
for early previews and feedback in the preparation of this document.</p>
<p><br /></p>
<p>Copyright 2019, 2021 Alexandre Oliva</p>
<p>Permission is granted to make and distribute verbatim copies of this
entire document worldwide without royalty, provided the copyright
notice, the document's official URL, and this permission notice are
preserved.</p>
<p><a href="http://www.fsfla.org/blogs/lxo/pub/0G">http://www.fsfla.org/blogs/lxo/pub/0G</a></p>