Re: A Credibility Use Case: Lessig v New York Times ("Clickbate Defamation")

Drew wrote:

> The only other scalable signal I think would be useful is if one could
> flag that part of the image is an original work being used out of the
> original context.

The challenge to this approach comes once we recognize that the authors of
content with "out-of-context" images are unlikely to be motivated to create
such flags since those flags' mere presence will tend to diminish the
authors' credibility, even if the out-of-context use is
completely legitimate. It seems to me that such flags will only be useful
to third-party content evaluators who have no beneficial interest in the
published content. Of course, third-parties are unlikely to convince
publishers to embed flags unless those flags are positive. Authors and
publishers will alway welcome positive credibility signals, but for
negative signals to be effective, there must be a way to create them
without the consent of their often unwilling subjects.

Certainly, third-parties are free to publish whatever they wish on
their own sites, and some sites already produce ClaimReview
<https://www.claimreviewproject.com/> or MediaReview
<https://www.claimreviewproject.com/mediareview> ratings that can express
both positive and negative content assessments. But, we need to ask: "How
will content consumers discover third-party hosted credibility signals or
claim reviews?"

It is useful that Google, and perhaps other search engines, sometimes
include claim reviews in search listings, since seeing them there will
either dissuade one from following a link, or at least alert one to a need
for more careful reading. However, as far as I know, we don't yet have any
browser extensions or features that allow users to discover relevant
signals or reviews while reading content. Ideally, when one is reading a
page, such as the one that offended Lessig, or one that includes the
Taliban flag-raising image, one's browser would indicate that relevant
signals exist and might allow one to open and read those signals. The
mechanism provided might be similar to what Hypothes.is
<https://web.hypothes.is/> already provides for annotations. Of course, we
might also define the means for encoding credibility signals and claim
reviews as W3C Web Annotations <https://www.w3.org/annotation/> and thus
leverage all the existing annotation systems.

We need also, I think, to address the issue of granularity. Lessig doesn't
challenge the body of the article whose headline offended him. I think he
also had and has no particular desire to challenge the general credibility
of the New York Times or of the article's author. He challenges only the
headline. Thus, he needs a mechanism to associate a signal with just that
headline and not with either the body of the article, its author, or the
NYT. The W3C Annotation specification provides a means to associate data
with fragments of a page and, if only extended to support credibility
signals and claim reviews, would have provided him the means needed to
explicitly tag just the headline, not the body, with his objections.

So, I agree with Drew that we need a means to "flag" parts of images and
other content. However, I think it unlikely that such flags will be very
useful unless it is possible for third-parties to create them, without
consent or cooperation of their subjects,  and for readers to easily
discover that such flags exist.  I think it would be useful for this group
to explore the mechanisms that can be used to do that.

bob wyman

On Mon, Aug 30, 2021 at 2:31 PM drew <drewwallace@gmail.com> wrote:

> Should the individual author own the credibility metrics, with the
> publisher being one of the inputs?
>
> So the NYT as a publisher would increcrement Ross Douthat's credibility
> score, but his byline/this article would carry a discrete, probably lower
> score than assigned to the NYT editorial board, alumni like Abramson,
> etc.... This would also allow op-eds to inherit a score from a known author.
>
> The only other scalable signal I think would be useful is if one could
> flag that part of the image is an original work being used out of the
> original context. This could warn viewers that memes, quotes, and image
> mosaics like the Isis Jima photo are derivatives.
>
> drew
>
>
> On Mon, Aug 30, 2021 at 11:01 AM Daniel Schwabe <dschwabe@gmail.com>
> wrote:
>
>>
>>
>> On 30 Aug 2021, at 02:29, <scsankaran@gmail.com> <scsankaran@gmail.com>
>> wrote:
>>
>> I feel this example, while valid, is peripheral to the central problem.
>>
>> Just numerically, the misinformation problem is to some degree about
>> journalists and readers of institutional journalism but to a vastly larger
>> degree about the millions of casual information broadcasters and consumers
>> on social media and messaging that form the information supply chain.  See
>> the recent example posts below that came into our platform.
>>
>> Sorry, I’m confused by your examples.
>>
>> Credibility signals need to surface the reasons for a consumer to be
>> anxious (or not) about these posts below at the point they encounter it.
>> These posts were seen by tens of thousands of people.
>>
>> The sort of credibility signals we’d care about to solve this problem are
>> very different than the ones being discussed in this email thread. This
>> can’t be a discussion about awards won, or anything you might use to rate
>> an established journalism organization.  Ratings such as those done by
>> NewsGuard, Media Bias Fact Check, etc begins to provide some value, but
>> barely so, in this situation.
>>
>> https://twitter.com/ClayTravis/status/1429174503930613764
>>
>> In the first one, is it being contested that the Taliban did not take
>> this picture, or that it is staged (is there any claim it is from an actual
>> fight?), or what are their intentions in releasing it?
>> According to Snopes, https://www.snopes.com/fact-check/taliban-iwo-jima/,
>> it was first released in July 27, in a post by apparent Taliban
>> spokesperson Hemat Mohammad along with a message that praised the Badri 313
>> unit's strength. This information could certainly be provided through
>> credibility signals.
>> Other than this,  it’s reuse at a later date, the apparent reference to
>> the famous Iwo Jima (staged) photo, and all the parallels and implications
>> that can be drawn between them, would fall beyond the scope of credibility
>> signals, no?
>>
>> https://twitter.com/SmartAssJen/status/1431793390224429056
>> <image002.png>
>>
>>
>> This post has several claims that evidently need to be better supported,
>> and fact checkers would do this, at least the quantitive one. The second
>> statement “…the same people…" is more  of a “ideological/moral” nature (The
>> other posts in the thread seem to confirm this perception).
>> Even if one could somehow verify the accuracy of them being the same
>> people (anti-vexers = supporters of criminalization of exposure to HIV), I
>> believe it would be less relevant to the actual (moral) message being
>> conveyed.
>> Beyond the provenance of the author, what credibility signals would
>> address this?
>>
>> Cheers
>> D
>>
>>