W3C home > Mailing lists > Public > semantic-web@w3.org > July 2019

Re: Deep Fakes, Phishing & Epistemological War - how we can help combat these.

From: adasal <adam.saltiel@gmail.com>
Date: Fri, 5 Jul 2019 19:07:55 +0200
Message-ID: <CANJ1O4r-85yt0ONTbYiuM8X2pSSD_Y8xbOijB_woO3SdAq0=Jw@mail.gmail.com>
To: Henry Story <henry.story@bblfish.net>
Cc: Paola Di Maio <paoladimaio10@gmail.com>, semantic-web <semantic-web@w3.org>
Yes. So two angles.
The “what is truth” angle and the UN of the internet.
I’m beginning to think the former may be easier.

Any luck with the latter?

Adam

On Fri, 5 Jul 2019 at 09:06, Henry Story <henry.story@bblfish.net> wrote:

>
>
> On 5 Jul 2019, at 05:37, Paola Di Maio <paoladimaio10@gmail.com> wrote:
>
> What I want to say Henry
>
> is that misrepresentation of truth is already widespread, and takes many
> forms
> (as in advertising and propagandas of all sorts)
>
>
> There are rules in the UK against false advertising, and in many other
> countries.
>
> If you analyze what adverts say, you will notice they rarely say something
> false,
> but they trie to build projections of how their product can help.
>
> Propaganda is also a form of projection into an idealized future.
>
> I’d need to study those.
>
> this technology makes it easier and can spread misrepresentation faster at
> mass
> information level, the underlying need for fact checking, understanding
> data bias,
>
> interpretation and context, and scratching beyond the surface of
> information are
>
> historical issues that exists well before this new capability.
>
>
> Each new technology requires new structures to be put in place to regulate
> their
> healthy use. These regulatory structures must come after the appearance of
> a technology, as it is difficult to legislate in anticipation of something
> that is new
> and so not quite known (and usually not even taken seriously, as only few
> are
> good at futurology).
>
> The problem with internet regulation is that it goes beyond the national
> and so
> lacks strong enforcement rules. After all the danger of nations imposing
> their
> regulations outside of their borders is that if everyone does this there
> will
> be conflicting laws, and so conflicting judgements, and so more and more
> reason for conflict.
>
> Instead of trying to go for one World Order, I suggest it would be better
> to
> make it possible for actors to make visible their ties to legal spaces
> under which
> they fall in order to allow good actors to distinguish themselves from
> those that
> do not want to make themselves responsible, and for these legal spaces to
> be
> diplomatically tied together in a web of nations, which could change as
> alliances
> change.
>
> PDM
>
> On Fri, Jul 5, 2019 at 11:20 AM Paola Di Maio <paoladimaio10@gmail.com>
> wrote:
>
>> Thanks Henry
>>
>> for the extensive reply
>>
>> >>>The mayhem appears when things
>> >>>are published as true by sites that look like official ones.
>>
>> But this happens already, quite a lot, everywhere
>> From political websites, to institutional, even scientific sources (a lot
>> of gibberish
>> that nobody understand and nobody can reproduce/verify, and noboy has the
>> time
>> to investigate further is published as science and taken as fact. Its
>> only when somone, often by chance stumbles across some issue that certain
>> things come to light)
>>
>> Official websites are full of lies or partial truths
>> Even omission of facts is a misrepresentation of truth
>> Telling the truth is actually perceived as a silly and stigmatized
>> (people are ridiculed when they things as they are, so there is great
>> fear in telling the truth) So I would say first and foremost is a
>> cultural thing
>> that we trust implicitly what information comes from institutions
>> But institutions have hidden agendas and use information not for
>> information sake and to make people more knowledgeable, but to influence
>> and stir
>> opinions and behaviours
>>
>> p
>>
>>
>>
>>
>> On Thu, Jul 4, 2019 at 10:09 PM Henry Story <henry.story@bblfish.net>
>> wrote:
>>
>>> On 4 Jul 2019, at 10:33, Paola Di Maio <paola.dimaio@gmail.com> wrote: >
>>> > Reality is so manipulated (at all levels) that humans have lost  (maybe
>>> > never had) the ability to understand of what is real beyond doubt,
>>>
>>> That is actually the subject of Epistemology. This comes in two parts.
>>> 1)
>>> The problm of definition: What is knowledge?  2) The sceptical problem:
>>> how can we know anything given that we can always find reason to doubt?
>>>
>>> Knowledge was defined by Socrates according to the reports by Plato as
>>> Justified True Belief. More than 2 thousand years later, after the
>>> development
>>> of modern quantified logic with Frege and Russell/Whitehead, the
>>> questions
>>> came to be to find logical necessary and sufficient definition of
>>> knowledge.
>>> These lead to well known problems defined by American Philosopher Edmund
>>> Gettier https://en.wikipedia.org/wiki/Gettier_problem
>>>
>>> Around the same time Modal Logic came to have a mathematical
>>> formalisation
>>> and Hintikka used this to defined
>>>  S knows that P iff
>>>    in all the worlds compatible with the information S has, P is true.
>>>
>>> Robert Nozick in the award Winning book "Philosophical Explanations"
>>> showed
>>> that there was a problem with this definitiion. By updating Descartes'
>>> Meditations to the Science Fiction realm, and arguing that we could
>>> always
>>> imagine that aliens from Alpha Centauri had come at night, kidnapped S,
>>> attached his brain to a super-alien-computer and induce in him fake by
>>> realistic
>>> sense impressions. Since this doubt can always be brought up in that
>>> form or
>>> the more ancient one of dreaming, the question becomes how we can know
>>> at
>>> all, since that possibility cannot be excluded.
>>>
>>> The answer come by way of using the David Lewis' later logic of
>>> counterfactuals
>>> that organises possible worlds by a distance relation. Redefining
>>> knowledge
>>> using counterfactuals as Nozick does, it turns out that one does not
>>> need
>>> to consider more distant and outrageous possible worlds to know some
>>> everyday
>>> fact about how much money one has in one’s pocket.
>>>
>>> I give an overview of that in "Epistemology in the Cloud - on Fake News
>>> and Digital Sovereignty" (And if you don't want to read the paper you
>>> will
>>> find two presentations with slides, one of which I gave at the Chaos
>>> Computer Club Vienna's Privacy Week)
>>> https://medium.com/@bblfish/epistemology-in-the-cloud-472fad4c8282 There
>>> I add a Cloud computing related twist to it, leading us to take seriously
>>> the locality of information.
>>>
>>> > The vastness of widespread deceit (about news, history, and even
>>> science!)
>>> > and limited resources to verify everything that we hear, we need to
>>> limit
>>> > our fact checking to the strictly necessary facts that support our
>>> > decision making/ So when I read or hear some fact, I do my best to
>>> verify
>>> > its true.
>>>
>>> Yes, so if you are going to verify the truth of a statement quickly you
>>> may need to use the internet to do so.
>>>
>>> In the pre-internet world, you would do so by finding someone
>>> knowledgeable
>>> on the subject, which in many case would be someone educated in the area,
>>> or working for a company that is known to be able to make knowledgeable
>>> statements on a topic. So you may go to a dentist to get a prescription
>>> for your tooth pain, or to get a tooth pulled, not to someone you just
>>> met
>>> in the bar, even if they can speak very convincingly on the subject. Or
>>> you could read a book published by an expert in the area, and that
>>> expertise
>>> would be verifiable by knowing which institution they were speaking from.
>>> Of course if you are a mathematician reading a mathematical proof you
>>> would
>>> just need to verify the proof for yourself, but you may yet want to
>>> filter
>>> the things you read by knowing where the person writing things came from.
>>>
>>> This thinking gets one to understand the role of institutions and legal
>>> systems in our claims to knowledge. To make statements in a factual
>>> context
>>> is to be make oneself responsible for what one says, and requires one to
>>> not follow up by saying something contradictory to that. To make a
>>> promise
>>> requires one to be able to follow up on it, and then to try to follow up,
>>> and so limits one's future possible lives to those compatible with one's
>>> promises. Entering an institution is to make a certain promise to uphold
>>> its values.
>>>
>>> But the web currently has not useful information about what institutions
>>> is behind a web site. A little typo, or clicking on a phishing link can
>>> make you end up on a web site that looks very much like what you are
>>> expecting but be a fake site. This was very unlikely to happen when
>>> buildings
>>> in a town gave you a way to recognise the institution you were talking
>>> to.
>>> That building would in any case mean the presence of people on legally
>>> delimted soil.
>>>
>>> So before the large public can even get around to fact checking we need
>>> to build an institutional Web of Trust (WoT), which can play the role of
>>> buildings in local life, by letting people know the legal framework a web
>>> site is tied to.  I describe how to do that in the blog post "Stopping
>>> (https) Phishing"
>>> https://medium.com/cybersoton/stopping-https-phishing-42226ca9e7d9
>>>
>>> This can be done with Linked Data because we do not require global
>>> consensus,
>>> and so we can allow different nations to have differnet points of views
>>> on each other and even how to map ontologies, when disagreements arise.
>>>
>>> > Deepfakes adds another layer to that manipulation and falsification of
>>> > reality, by leveraging new technology.
>>> > I see two areas of concern
>>> >  a) technology ethics - a fun  technology developed
>>> > to animate fictional output is used to falsify reality  (making people
>>> say
>>> > what they have not) with potentially devastanting consequences is  not
>>> > entirely new-manipulation has always occurred by twisting, falsifying
>>> > or taking out of context what people may say.  Misinformation and
>>> > misrepresentation are  a less technologically sophisticated, but with
>>> > similar consequences (to manipulate public opinion and behaviours)
>>> This
>>> > already happened with emails.  Deepfakes is a progression of  spoofing
>>> > tech where someone fakes another person email address.
>>>
>>> Deep fakes are not a problem if they are annotated as fictional.
>>> Terminator
>>> 1, 2 and 3 did not cause global mayhem, because they appeared in cinemas
>>> and were clearly labled as science-fiction.  The mayhem appears when
>>> things
>>> are published as true by sites that look like official ones.
>>>
>>> > b) the increased value of authenticity, and authentication tech
>>>
>>> That will be important especially for allowing private citizens to also
>>> make clear which legal space they are speaking from, when say they
>>> publish
>>> a photo or film about something happening.
>>>
>>> > From a systems view point, another layer of risk, can be addressed
>>> > with  another layer of architecture (strenghten authentication layer?)
>>>
>>> Yes, we need a new layer, but not the authentication one. We have that
>>> already. The domain name to DNS authentiation layer technology does
>>> its job well enough if one uses X509 certificates and DANE on DNS-SEC.
>>>
>>> What is missing is the institutional web of trust that can then be used
>>> by the
>>> browser to display rich information on a secured screen such as the Apple
>>> Touch Bar, in a seamless but helpful way. The information contained
>>> in X509 Certificates is much much too poor to be of interest and hence
>>> of use.
>>>
>>> For an example of how this institutional web of trust could be tied to
>>> hardware see the blog post "Phishing in Context - Epistemology of the
>>> Screen" https://medium.com/cybersoton/phishing-in-context-9c84ca451314
>>>
>>> As for authentication of citizens using Verifiable Claims so that they
>>> too can
>>> make claims (such as location claims if they were a witness to something)
>>> needs the institutional web of trust to work for networks that go beyond
>>> a few degrees of seperation, since if you go a few more jumps you have
>>> the
>>> whole world in your network.
>>>
>>> Henry Story
>>>
>>>
>>>
>>>
>
Received on Friday, 5 July 2019 17:08:31 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:42:07 UTC