credweb

Minutes: Credible Web CG (27 June 2018)

Topics

Resolutions

Minutes

Admin

Sandro: Agenda https://credweb.org/agenda/20180627.html .. Pretty open agenda - probably short meeting, unless others bring ideas .. let’s try recording this meeting, and then I’ll scribe from that. .. I’ll keep the recording confidential unless I get permission from each person who speaks .. does everyone agree to me recordning this meeting? .. if you don’t agree, please speak up or leave the meeting

Everyone present assents

Sandro: so lets dispense with the scribing .. no minutes ready to approve .. any new agenda items?

Intros

Sandro: seems like everyone knows each other on the call today .. let’s skip intros

Event Reports

Sandro: I was at Global Fact V last week, in Rome, and the agenda includes links to some good reports on that. .. Most professional fact checkers. An honor to meet some of them. But for me .. the main thing was talking to the tech folks behind the fact checking operations.

ClaimReview

long digression from event report

.. Following some discussion there, Dan Brickley made a github issue trying to summarize all the open issues .. on ClaimReview. .. https://github.com/schemaorg/schemaorg/issues/1969 .. Jon did you see that summary? Jon: He pinged Rob Sanderson and me and we had some responses to that Ed: I’m also very interested in that Sandro: Kind of a meta point, but it seems like the folks who’ve developed ClaimReview to the point it’s at feel some ownership, and don’t want to just move it to W3C, which might feel like other people would be taking it over and maybe messing things up. .. I’, suggesting they’ll probably want wider industry buy-in, and W3C’s usually a good way to do that, but we’ll see Jon: I feel like there’s a disconnect between practice and theory here .. which can happen around standards .. where the application falls by the wayside .. I’m at a bit of a loss .. I’ve identified a situation that doesn’t require any additional spec work, .. it’s about getting what’s there applied in a more effective way .. that will deliver the results that everyone says they want .. but I feel like I’m hitting a brick wall Sandro: I think it’s just about people being too busy. I think .. they agree with what you’re saying, that there are some fairly simple .. problems, but no one’s quite found the time to sort them out yet .. The people who work in Fact Checking orgs, and the tech people behind them, .. have this really important mission and it’s hard to find the time to .. work on infrastructure like this. Ed: I agree it’s hard, it’s going to need some incentives being introduced Jon: But there’s a huge incentive from the platforms, .. there’s search engine love being left on the table Sandro: Also FB announced last week they’ll be consuming ClaimReview as well Jon: I can’t imagine a bigger incentive for a media organization or .. fact checking org. You do the fact check work, write it up, and .. then you fill out this form. How do you not want it to be the case .. that when someone lands on the page you reviewed, in search results, .. they know that you reviewed it? Right? That’s the whole point of this. .. And the feedback loop is not getting connected for a very simple reason? Scott: What’s the reason? Jon: For most of the claimreview data that’s being pumped into the ecosystem .. the URL of the target of the review is not included. So the ecosystem .. does not know. In google search results, if the Snopes article itself is .. the thing that shows up in the results, then Google paints a label on it. .. And says this is a fact check, and it’s 2 stars or whatever. But it’s just .. a tautology, because you’re on the fact check. What you want to be notified .. of is, if I encounter the Breitbart article or whatever it was, in the wild, .. the search engines want to know! There’s been fact checking effort directed .. at this resource that we can now surface. We can put the label on that! I think .. that’s what everyone wants, right? Sandro: So, does any platform implement that? Jon: They don’t have the information to do it. If Snopes doesn’t say .. in their claimreview markup that they’re talking about some particular .. article, then the platforms can’t know it. Sandro: It sounds like the bootstrapping catch-22, where platforms aren’t .. using it because it’s not there, and people aren’t puting it there because .. it wouldn’t be used if they did. Jon: I don’t know if the second part of that is true. I spoke to someone .. from Bing at WEF and he said he wished they were getting it, but they’re .. not getting it. Ting: That’s definitely the case. I’ve talked to Google folks and we would .. all love that. That’s where the search results would have the most impact. .. We’re making a proposal to change the schema Sandro: Are you doing that on the github issue? Ting: I’m not sure, it was submitted to schema.org, Dan Brickley Jon: That’s really interesting. I wonder if you have observed or heard about .. the following schenario: However the spec says you should report the target, .. that’s a field that needs to be filled in, and the people in the fact .. checking orgs may not understand what that’s for or whether to do it. That’s .. unrelated to the schema. Ting: Yeah Jon: And in at least two cases I’ve talked to, in these orgs, that’s one extra .. step they don’t want to do. Ting: In the original schema, there’s a field where folks are supposed to .. put the original article information, but it was not properly used Sandro: One thing that might help, which I saw demoed but may not be .. released yet, it’s a web form you fill out, and it goes straight into .. their claimreview database, which is of course a little worrying, but .. it also goes into the open data on datacommons.org. The theory is that .. UI will help people get it right. I think about half the claimreview .. markup right now comes from one widget – from Duke Tech & Check – .. so maybe if that were changed it could help. Chris Guess would be the .. person to talk to about that. .. All in all, Jon, I think this is a real problem and people would like to .. solve it, we just need to get the right people talking to the right people .. with the right motivations in place Jon: And is this group the place to solve it? Is that part of our charter? Sandro: To try, yes. But if the key stakeholders don’t want to come here to .. talk about it, that’s their choice. Scott: Jon, I would love to help in some way. It seems like a solveable problem. .. I share your frustration. Maybe we can team up a bit. Sandro: one of the things I thought would be more helpful, was if I could .. say: come to the CredWeb meetings because we have lots of motivated and .. skilled people who can help. But, there aren’t a lot of people coming to meetings .. who are able to work to help claimreview, I think. Without a lot of .. resources, it’s harder to make the case. Ed: I’ve been on some calls coming out of Tech & Check, and it’s clear .. these folks are very busy. And they’re focused more on discovery. I .. told Mevan I would take a stab at a peice of it, and I’ve been working a .. bit on that. I don’t think their participation or lack of participation .. in credweb is a reason for credweb to not take on claims and standards around .. claimreview. I think this is the most actionable work we can do, so I would .. strongly recommend this group not move away from taking this on. Let’s .. figure out how to help .. the folks really busy with their day job. Sandro: I think github is the place for folks to work on claimreview, and .. schema.org in general. There may also be f2f meetings at some time. Jon: Do you feel like the tech folks in the fact checking orgs are paying .. attention to that github Sandro: I think, at least a bit. I think Dan is counting on them doing so. .. That was a long digression. .. Next steps? Jon: Can you point me at FB consuming this, and the Google tool? Sandro: I’ll see what I can dig up .. anything else on ClaimReview Ting: Do we know if FB will publish ClaimReview, based on the fact checks .. they gather? Sandro: That’s a good question. Reading between the lines, I though I heard .. that was likely, but I don’t think it was promised. I heard the comment from .. FB that they want to simplify the work for Fact Checkers as much as possible, .. so not make them input their data into two different forms. So just input .. into the Google single form, or FB single form, the Duke single form, and it all .. ends up everywhere. .. One story I heard, not sure I have the details right, in South America, .. is some FC org factchecked something, so FB demoted some content, then .. the folks who had their content demoted sued the fact checkers. .. So the FC asked the question: will FB fund a legal defense fund for .. fact checkers? Scott: What was the answer? Sandro: essentially, “We’ll look into it”

Calendar

Sandro: We settled date and location for July F2F meeting .. thank you Ed for finding space .. thank you Scott for finding backup space .. see link for registering or regrets .. Looking at the next few weeks .. no July 4 meeting .. then two weeks of regular Wednesday meetings .. then cancel July 25th as day before F2F .. for Monday meetings, I’ll have to talk to An when she’s back from China .. Note MisInfoCon DC Aug 6-7 .. W3C technical plenary in Lyon in October .. they gave us a room on Friday, but that conflicts with JSON-LD .. so I’ve asked if we can switch to Monday/Tuesday .. and it depends on room size. .. so don’t book tickets until we figure this out! .. anyone know they’re going to Tech Plenary? Davide: +1 Scott: kind of conflict with JTI meeting in France a few weeks earlier

Charter Progress

# Ting Cai says: : Will Facebook publish ClaimReview metadata for their own fact-checking efforts?