Minutes for GAIA at IETF-96
minutes-96-gaia-1

Preliminary Agenda
Global Access to the Internet for All
GAIARG Meeting @ IETF-96
Berlin, Germany
Thursday, July 21, 2016 (CEST)
14:00-16:00   Thursday Afternoon session I
Location: Schoeneberg
Audio: http://ietf96streaming.dnsalias.net/ietf/ietf968.m3u
Video: http://www.meetecho.com/ietf96/gaia
Etherpad: http://etherpad.tools.ietf.org:9000/p/notes-ietf-96-gaia
Jabber: gaia@jabber.ietf.org

Minute taker: Lucas Jenß <lucas.jenss@gmail.com>
Minute polisher: Mat Ford <ford@isoc.org>

Welcome, Agenda Bashing, Minutes taker, Blue sheets, Status
  Chairs
  10 mins

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-2.pdf

Refugee Hotspot - providing Internet connectivity to Refugees and Migrants
  Niels ten Oever & Shane Kerr
  20 mins

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-0.pdf

[...]
Niels ten Oever (nto) (paraphrased): While we did use Debian becasue that is
what we were used to, we would like to be educated on OpenWrt usage, which to
use seemed to be kind of a pain [...] Shane Kerr (skr): Work available on
GitHub, help appreciated [paraphrased] (at https://github.com/RefugeeHotspot)

<Q&A>

?: What does your device do? Where does it get Internet access from?
skr: 3G/4G Internet access, and GSM based network and then transit to wifi
?: benefit over using hotspot on a phone is traffic shaping?
skr: that is one benefit
nto: also ppl don't have financial means to get a sim, so in this way we can
serve a community instead of handing out individual sims Arjuna Sathiaseelan
(chair): how many hotspots are out? nto: one is in the wild, another is going
out next week, we got some funding for 10.. ppl will be able to buy it via the
website, and buy one give one, but have not looked at bigger funding proposals
because we are still exploring what the community actually needs. could do
bigger battery, smaller battery, waterproof casing - need to see what people
are most interested in. JULIUSZ CHROBOCZEK (jck) : i am impressed that you are
checking what your users need rather than just hacking lots of things that
don't fit user expectations. agree that openwrt is a complete pain to set up
and hard to debug. but it is done by good specialists and when things work it
does most of the time everything you need out of the box. there are quite a few
openwrt people here this week, and rate limiting people are here too - they
will tell you that rate limiting is very difficult - 'wifi sucks' is their
slogan - these are difficult problems - use their expertise rather than trying
to relearn and reinvent things. you mentioned preferance for raspberry pi style
hardware - agree - we need to work on generic debian boards. there is a
community interested on working on this. Roger Baig (rbg): what is the degree
of coordination with freifunk people? afaik they also work with refugees and
providing internet access to them. also related to community networks and
raspberry pi - this was a bit shocking to me - there are a lot of openwrt
projects that provide hotspots and also mesh out of the box etc. nto: we've
tried to coordinate w/ groups in amsterdam, we're in contact with refugee
collective, but we should be in contact w/ ppl providing them connectivity as
well - and freifunk is definitely high on priority list. skr: I confess we
didnt do a months long review of all possible hardware options. some of the
stuff mentioned is out of scope for our Minimum Viable Product. mesh networking
is cool, but out of scope. e.g. we don't have IPv6 setup, but we do use dnssec.
don hollander: there's a group in NZ called 2020 communication trust
2020.org.nz, they do work with refugees in NZ. Beyond technology, they provide
training and take 2nd hand computers, refurbish them and make them available to
refugees. nto: In our experience, if you provide ppl with hardware you also
need to provide support, so thats out of our time availability unfortunately.
something else that is nice - because of this project people have been talking
to us and the centre in the netherlands that is responsible for distributing
unemployment benefit is refurbishing their network and so we are going to
receive 400 old wifi access points and seeing if we can get openwrt working on
those and continue hacking. skr: gave a similar presentation at the RIPE
meeting, many ppl approached me about orgs with related goals. lot of work
going on in this area - surprised and pleased to learn that there was so much
interest in and support for this work.

</Q&A>

The dual role of community networks: Internet access vs. local services
  Panayotis Antoniadis
  20 mins

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-3.pdf

<Q&A>

Mat Ford (chair): As you were presenting I was thinking an obvious obstacle is
that as a user you have to connect to the network. I have to do something to
connect to a local service. Have you done much thinking or work about making
these simple technical hurdles more seamless? Or are you saying that's the
point and you want to have users have a cognitive interruption. Panayotis
Antoniadis (pas): Can make local access networks that also give access to the
Internet. Local networks are interesting because they are tangible, can be
complemented with people, posters, visual cues. Big advantage: Nobody has to
subscribe to them beforehand, its a decision of the moment if you want to
connect or not. If motivation is strong enough, in principle everyone can
connect. Privacy issues of location based services with global services are
important. al cano santana (acs): Volunteer for guifi.net administrating
servers providing local services for people. (on slide extrovert/introvert) ->
is this based on some social study, or is this a graphic that is just a general
idea? pas: I would like to make it accurate if you can help me. This is a
network in Greece - a big access network, everyone can connect - other extreme
is a network that you have to buy an antenna to be a part of. acs: do you have
any data related to this? i am using data on all wifi community networks to
understand what people are doing around the world. pas: no - i'd love some help
- maybe we can collaborate on this. gio mazzurco (gmo): (libremesh) sometimes
we build community networks, and they are used to access facebook, which is not
what we want to happen. so we need a set of applications that take advantage of
the local network. that seems to be an empty space. it seems to us that
companies are not interested, generalising scope of libremesh (librestack) to
fill gap - interested to collaborate with your organization to fill that gap
pas: of course. you mentioned needs - i always talk about this - people ask if
people don't need it what is the point, but marketing creates needs - people
sometimes need to be lead. it's a dilemma - should we strive to create
opportunities for people to go to places they didn't know they wanted? jck:
when you were mentioning the garden, you said that they were much more open to
letting people in from outside when you said that they could prevent ppl from
accessing facebook. are you advocating for networks where the operator can
control the sites accessed? pas: no, I would advocate for local networks w/o
access to the internet, and encourage ppl to use them. for example captive
portals in train stations - this is my design space - what could be there, what
should be there? if the Internet is not available that's fine. i don't think
every space should have Internet in order to offer a local service. jck: think
its a slippery slope guido iribarren (gin): in my experience, I saw a
difference between making a local service also available over internet which
takes time, if network has Internet connection, then it adds value to local
service for example if its some xmpp server to take the time to make proper DNS
records and transparent for user if its on local network. Do you agree that
this project has this vision as well, or is it only focused on local services
and not thinking of internet? pas: insistence on local is just to create a
balance, if the internet is avaialble and we can use it we should use it. if
things are local, we should be creative in thinking of local services.
intermediate stages are also possible. different cases imaginable, open to all
combinations, just starting from this extreme stance to make a case that ppl
can have a local network and can do whatever they want with it. we are not
religious, we want to create options. sometimes the internet does not exist.
its crazy that right now, when the internet falls, we cannot send an alert to
our neighbours, from an engineering perspective we should have such networks
for backup. jose saldana: these days [in spain] ppl are crazy with their
pokemon game (go). could this match with your idea of a local thing? game
attracts people to a physical location, where local network might be present.
is this a potential use case? pas: my reaction is negative to this addictive
games, but of course if people want it. I can imagine ppl running this on local
servers, for speed, resilience. but it depnends who designs them for what
purpose. local network creates a political negotiation process - what network
should do with data and so on. crazy that location based things are managed by
unknown remote entities.

</Q&A>

FreeSurf: Application-Centric Wireless Access
  Zhen Cao
  20 mins

* slide 11
jck: are you saying that eduroam does triangular routing for all user data?
zhen cao (zco): for the authentication data, not for the user data

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-4.pdf
https://github.com/freesurf (apparently empty at the time)

<Q&A>

jehan tremback (jtk): don't you think that censored internet access will be
rejected by users? we had walled garden service providers, e.g. compuserve, but
they went out of business. people use facebook a lot, but they like to go other
places too. zco: good comment. for big guys that are rich enough they can just
offer full Internet access. then that's no problem for your case. justin dean
(jdn): I really dislike this whole idea. its hard to argue with giving
something for free, but it's kind of carrot and stick - people get used to
getting something for free so they don't want to pay for it and then find they
can't get access to something else because there isn't the critical mass of
users needed to pay for it. I could see someone like ? buying services
everywhere, and then I'm stuck with their crappy platform instead of using pied
piper. zco: can agree with you, but for these guys who want to give access to
their users, i think they are good guys compared to operators who stand in
their way. freesurf can be complementary to full Internet. gmo: I feel offended
by the naming of the project, because freesurf has nothing to do with freedom,
and nothing to do with global internet access for all. zco: don't agree because
this doesn't say that you can't surf the internet with your provider. the
policy defined here can allow for full internet access [...] probably you
missed this in my talk. gmo: to have access to the network, you have to login
through a big player zco: yes, for authentication, but after authentication you
can get full internet access, not just access to the service provider's domain.
gmo: by default and by design, you're excluding anyone who is not using a big
service like facebook zco: this is just an example. it doesn't have to be paid
for by facebook for example. mat ford (chair): so a service provider could
offer free access because it's subsidised by whatever commercial concerns
right? zco: the architecture itself is neutral to these business
considerations, but it allows you to provide full Internet for the people you
want to support mat ford: there's no such thing as a free lunch - someone has
to pay for the infrastructure - I'm just trying to understand who's paying for
it, and how, and why. zco: there's no free lunch, but there's a good tradeoff
jeff osborn: i'm finding it hard to find a slice of this to defend. use of term
'free' and 'surf' is probably not helping. most of us got onto Internet by some
method that didn't cost that much and we were able to get to everything. this
could be applicable to work i'm doing in guatemala and in ghana where farmers
who have $3 communications devices and a budget of $1 a month to communicate
with them would love to figure out a better way to get weather services and
crop prices in advance of a yield. this is an interesting idea, but the bad
thing is call it free, or internet, or surfing. zco: it's complementary right -
provide a more cheap device for users and cheaper internet for everybody.
that's for the daily part. you can't take your broadband anywhere. jeff osborn:
it's all hard, but this is interesting. christian o'flaherty: i have a
clarifying question. what you're offering is sponsored access? so as along as
you have an account with the sponsors e.g. amazon, you can access the internet
fully? so its a subsidized model for community networks if they would like to
add that model to their authentication? optionally they could receive money
from third parties by this route. is this correct? zco: yes. al cano santana:
regarding the critics, this is my first time here in an ietf meeting, so i have
a question for the organisers, there's another project in another room which
didn't get these critics. so i want to ask what are the admission criteria for
presenting in this group? mat ford (mfd): the bar is very low, i email the list
asking if people have presentations al santana: so for you this presentation is
OK? mfd: i think we learned a lot from this presentation. the presenter has had
hopefully useful feedback. as chairs we've learned we should maybe pay more
attention to what people are planning to present before we accept them but I
don't think this presentation was completely out of scope.

</Q&A>

Sustainability in guifi.net: Initial results of cost sharing
  Roger Baig
  20 mins

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-5.pdf

<Q&A>

jck: what is the scale? you gave us two areas with 10,000 nodes but what is the
total size? rbg: close to half of guifi.net - so 15000 or 20000 nodes - we also
discovered a lot of hidden nodes - officially guifi.net has ~30k nodes but
professionals were deploying infrastructure without declaring it - we
discovered that there are more active nodes than are declared on the website
jck: so you're aligning incentives christian o'flaherty: some volunteers, some
professionals, some common pool? rbg: no, common pool resource is network
infrastructure. infrastructure means nodes and links. some contributed by
volunteers, some by public administrations, some contributed by professionals.
customers get access to common pool through professionals. in other community
networks this is not allowed and it ends up being all volunteer. christian
o'flaherty: do you know the percentage of nodes per player? rbg: this is
changing. at the beginning guifi.net was almost all volunteer based in 2004. in
2005/6 the first professionals appeared, so those who wanted to be connected
but didn't want to climb on the roof paid someone to do the work, afterwards
these pros and others started offering services, e.g. internet access, by a
proxy. moving from 100% to 80% of professionals. jehan tremback: major
difference between guifi and major isp is governance and corporate structure?
rbg: yes, which has a lot of implications. like you can be a part of the
governance structure. guido iribarren: how do you monitor that providers don't
over report their investment. rbg: reports are made publically and other
providers do a good job of policing the truthfulness of each others statements.

</Q&A>

Wi-5: Advanced Features for Low-cost Wi-Fi APs
  Jose Saldana, University of Zaragoza.
  5 mins + Q&A

https://www.ietf.org/proceedings/96/slides/slides-96-gaia-1.pdf
https://github.com/Wi5
http://www.wi5.eu

jsa: *jokingly?* directed at nto and skr "after about a year and a half of
openwrt, we more or less have it in check" [laughter :D]

jck: you still need to reassociate, you just don't need to renumber
jsa: layer 3 is not aware of that - it's always hearing the same MAC - so there
is also here a dhcp server and the controller. [moves to next slide to explain].

mat ford: [out of time]
mat ford: please send any feedback or suggestions to the list, one contender is
a document about deployment experiences. if you'd like to contribute let us
know.