Notes Human Rights Protocol Considerations (hrpc) RG
https://datatracker.ietf.org/rg/hrpc/about/
IETF 116, Yokohama
2023-03-31, 0930-1200 local time
Sofia Celi and Mallory Knodel, co-chairs
Agenda:
Hacks, Leads, and Revelations - book presentation: a hands on guide to
analyzing hacked and leaked data
Presents various tools, datasets, and skills that journalists can
develop to do analysis of hacked or leaked data sets, like blueset,
aleph, and python programming. also made their own software, blue leaks
explorer, to make it a bit easier to query it.
https://nostarch.com/hacks-leaks-and-revelations
[Presentation]
Q: curious, if you have chapter on ethical considerations re: the links,
and how students etc should deal with that?
A: yes - different levels of ethical considerations, especially around
source protection. PII need to be careful around, not publish if you
dont have to.
Q: do you have a chapter on how people to encourage to produce
datasets, your examples are from the US but these types of leaks dont
happen everywhere?
A: lots of leaks around the world, this is the stuff that Micah knows
the best, hence that focus. But they have seen leaks from eastern europe
and other places. Hopes that if more people starting digging into what
is available that will encourage more leaks happen.
Q: Your exmaples are about dumps and leaks, since we are here at the
IETF, do you have examples of network traces?
A: There is no work on PCAP, there is an appendix on webscraping that
shows you how to use beautifulsoup and cellenium to gather your own
datasets and scrape them yourself. More active stuff gets into legal
grey areas, so probably consult your lawyers
Q: Do you consider protecting against stenographic techniques to
protect sources?
A: The most secure way to protect your sources is not publishing, but
that needs to be balanced to transparency for your readers.
If you do want to publish, you can quote from it. If you want to
publish documents, you can recreate it. For instance in audio, you can
re-record it. So if there are hidden watermarks that won't be included.
So there is a whole range of things.
Reframing cryptography for activism, in 7 parts: introduction, protocol
design paradigm shift, definition of grassroots organizing, lessons from
history, lessons from the current landscape, tlgro: trust infrastructure
for grassroots organizing, conclusions.
Leah was focusing on number of properties of security, making sure that
needs are contextualized and focused on offline trust - which is key
given online info glut. Low-tech fallbacks matter, as does the role of
human trust which they are applying to a grounded trust method.
The choices we make in the design hold power and we want to communicate
about - who does our work serve? who holds power? where does the trust
come from? who has meaningful choices?
Q (Shivan Shahib): Thank you for the talk and this work. Have there been
groups that are working with this trust infrastructure?
A: Not yet, we are finishing the security analysis. I want to be very
careful, implementation is important.
Q (Mallory Knodel): I hope you consider hrpc as a place where you can
do this work. Can you talk a bit more about the annotations. Are these
visible or is it just a hash?
A: The annotations are not public, but only available for those that it
is encrypted to. We're not trying to fix physical risk trough digital
technologies, this is impossible.
Q (Erik Nordmark): Thank you for the talk - many of the tools we have
can be used for this. In a different time and world transparency and
visibility is one way of shining a light, how do you ensure that this is
not used by hategroups? Is there any work going on in that space?
A: That is something to continue to think about, with any open source
tool this is an issue. In facilitating technology trainings we are
explicitly aiming at grassroot movements. People who work on this
project, will likely be working with marginalized groups and not take
the tools to help hate groups.
Q follow up: as we can see with social media, we can see that these
platforms are being used for hate. So can we tilt the playingfield to
make it make it harder for bad actors to use these tools.
A: yes, its a hard problem. Other people are working on this.
Q (Lars Eggert): remembers PGP, this feels like PGP on steroids, web of
trust concept has natural scaling boundaries, where do you scale to? a
100 people? a country? etc
A: web of trust, there is a lot of paralels. but his has no PKI, but
the second difference is the scale, this is community based platform, we
are operating on a communinity basis, the scale is different - there
will be most localized.
Q (Lars Eggert): we had PGP signing ceremonies at the IETF - but then
they would never meet again. We can establish some trust, but how to use
these technologies if never meet again?
Q (Niels ten Oever): How do you do discoverability? how do i know if
there is tigro object of a meeting?
A: Tigro is not its own platform, it takes content from social media,
still playing with this. You take the content on social media (event
post,image) and hash it, so you have a short identifier that you can
upload into the tigro server - and then you can see if anyone else has
uploaded info about that same hash.
Q: what if the governments forces shutdowns? what about offline
functionalities?
A: thinking about decentralizing this, encrypted version firechat
(insure way of bluetooth),. Would be possible technical, have to think
more about the trust component of that.
Q (Lars Eggert): There was a delary tolerat research group, that is now
a WG, which might be relevant
Q (Gareth Tyson): How do you account for centralization in networks
producing vulnerabilities because of dependency on a centeal point?
A: Data about centrality is not shared through the system.
Q (Gareth Tyson): Can you produce indirect trust relationships?
A: Yes, we are working on this, but a lot of user studies are needed.
Ekr: some comments have been addressed, some of have not.
Mallory: folks were clear about what they took on board, and why they
did not include everything else. This is a research group.
Collin: did not realize that it was progressed back, we do need
consensus from the group. we do need the people who commented to feel
like their comments to be addressed by the group. we need more
discussion on the list.
Niels: so lets discuss some of these issues, it will save bits and
bytes.
Mallory: what is the date of the last message?
Ekr: bit about content agnotiscim is bad: content agnosticism is not a
property of the protocol but of something else. We need to clear up
things that are in the protocol vs other parts of the system.
Collin: people need to sit down and discuss.
Mallory: my sense is that the authors explained why they did not want to
include some things and then silence.
Collin: you need people to respond to the draft authors. We need a
response from the people sending the comments, they either agree, have
further comments, or agree to disagree. I did not see the closing of the
list. There is no complete agreement.
Mallory: talk to niels, not me. I am not the author, i am the sheppards.
Ekr: last message from me is march 15th, and i do not feel my comments
are addressed. there are still open issues.
Niels: okay - so lets sit down and do it together.
Mallory: happy people want to commit to sit down. But the rest of the
group needs to weigh in, it is not just Ekr. there might be moments
where the authors do not incorporate everything, me as the sheppard, i
am happy with that. Collin can you nod? he's coming to the mic, its not
fine.
Collin: you need to get confirmation from the folks making the comments
that the suggestions are addressed. You cannot make a consensus call
based on silence.
Mallory:is that person ekr?
Collin: yes
Mallory: i am not asking for a time-out, i am asking the authors + ekr
to resolve this within a reasonable amount of time.
Sofia: other people on the queu
Seb: what are these documents for, what impact do they have? whats the
goal?
Niels: goal is to make the doc go to an informational doc. its a
normative output documents.
Shivan: process question - does this need consensus? what does that mean
in the context of an RG? I thought RGs did not need consensus.
Collin: IRTF docs do not need consensus. But the group has decided to
make it a consensus draft. You can also make it non-consensus, but then
you have to change the boilerplate language when it goes to the editor.
Mallory: we could reconsider the status of the document.
Niels: I am happy to work with ekr, and want to respect the work of the
people who contributed to it.
Collin: consensus or not, if there are technical flaws those should be
addressed.
Niels: lets take that of
list.
Mallory: draft-association: waiting for doc sheppard. its now in the
datatracker.
Collin: data tracker does not notify people, so please email them as
well to let them know.
Still being developed, have a look at the current structure. Please add
examples and comments.
Q (andrew campling): observation that some of the recommendations could
be negative in the context of CSAM. a lot of CSAM is 'self-generated'
and the tools that might prevent that, will be undermined by the recs
you make.
A mallory: we are focusing on the harm, not imagines of the harm.
Q (andrew campling): CSAM is also about grooming, which is real harm.
Juliana: we do not focus on grooming.
Seb: thanks to tom ristenbark (?) wondering what the norms are around
referencing in the ietf / irtf?
Sofia: folks from Tom's have presented in the group. We are also working
on an academic paper on his work.
Niels: would be great if this would be a research group document, brings
in great experts like Leonie Tanczer. Happy to help.
Sofia: let us know who ele we should bring in.
Shivan: the threat here is the intimate attacker, we could generalize
from IPV to any considerations that might have access to your device,
like immigraiton officials and your employer.
Sofia: we focus on IPV, but you see the intimate access in other
contexts - we could expand it to other forms.
Mallory: at CDT, we work on bossware and online surveillance of kids -
all these things have common elements. We can look and analyze the
software itself
Collin: important topic, a good draft for this group to be working on.
Following on from the previous comments, i would be cautios about trying
to broaden it too much. Focused drafts are easier to complete and more
impactful.
Sofia:
https://datatracker.ietf.org/doc/html/draft-celi-irtf-hrpc-ipvc-00
All other WG business - remainder
Mallory: charter, i dont have slides. Want to change the document around
the words guidelines, as irtf can't do prescribe.
Colin: Mallory and I agreed yesterday, we have small text changes left
but we agreed on the direction, we're almost there.
John Curran: We should probably redefine what 'addressed' means, it is
probably perhaps 'considered' is enough.
Mark: I support the recharter and I think it will make the work
stronger.
Mallory: You might be interested
https://datatracker.ietf.org/doc/html/draft-irtf-hrpc-political as well!
Avri: With the recharter, would implementations fall into scope now?
Mallory: That would be interesting to me, perhaps you can make a
suggestion?
Minutes by Corinne Cath and Sofía Celi