Tools, techniques and limitations
Login to save
07:30 PM – 09:00 PM IST
Social media platforms have been at the forefront of the conversation regarding regulation of fake news, mostly for being slow responders, and thereby allowing multitude of objectionable content to remain online. What role, in the opinion of the speakers, should these platforms play in facilitating and helping the fact checking community, and making the process seamless?
Thanks Torsha. Is your question directed at someone in particular? Or should we ask all the speakers to respond?
I think, given the diversity of the speakers, it may be intersting to hear all the speakers? Though I leave it to you if doing so affects the internal time limits set.
No problem. We will have the speakers also respond here in writing.
I agree that platforms tend to “balance” and drag feet on fake news, and it is widely acknowledged that their economic incentives are aligned to more activity, more eyeballs, more ad revenue, higher valuation. This ignores the social consequences and damage to democratic values and they do not form part of equation for decisions. Platforms can, at transactional level, make costs of peddling fake news higher (eg by referring appropriate cases to Law enforcement agencies, proactively providing evidence) . Mere deleting posts, or suspending accounts is not deterrent for habitual defaulters. But more meaningfully, I reckon, autonomous governance boards, of the kind facebook has instituted last week, will make a difference. Governance board comprising reputed professionals and free of any conflict of interest with business of platform, should be more empowered than they curently are. That seems the way to go. Fact checkers should get an autonomous, transparent and quick complaints/appeals forum in these governance boards. Its early days, but policy advocacy groups and public at large, must get these governance mechanisms to be truly autonomous, more powerful and accountable.
Given that platforms are driven by engagement, and fake news also represents engagement, platforms have an incentive NOT to delete/suppress fake news. Maybe large platforms like Facebook will do it to build credibility, because they have sufficient engagement not to care any more. But otherwise it is a tradeoff between credibility and engagement
I often see people using the #hashtag that is publising the fake news in their counter post. Dosent that boost the fake news? and if so, what’s the best way to handle it?
Shared. Will also get a written response here.
Question from audience: Elaborate on the human aspects of fact checking. How does the team at AltNews go about verifying the authenticity of content?
What are the challenges - with respect to costs and resources - that The News Minute has to face in collating, filtering and disbursing fake news?
Question for Anna and The News Minute: while all this fact checking is great, how far does it travel on the ground? For instance, does it stay only on Twitter whereas the fake news goes offline via other media?
Question for Pratik: What are your views on open source communities being able to help with tools to combat fake news. Will AltNews be open sourcing the platfom and tools they’ve built, in the future?
Shared your question.
I just wanted to give my 2 cents on the first part of your question. I think that there are also a lot of fact checking adjacent tasks that could use some automation, community effort and new tech development or just minor tech support.
1. Collection of data from these ‘closed’ platforms. If you have a large pool of people that could crowdsource the type of stuff they see circulating on their family whatsapp groups and instead of directly bombarding the fact checkers with requests, provide them with a continous overview of things like “most requested fact check”, “trending right now” etc within that group
2. annotation and translation. from a research pov, a lot of the data that has been collected could still use human effort to clean up and annotate to make its exploration easier. Also annotating data in a meaningful way that feeds development of other ML applications
3. data disemmination. I think an undervalued aspect of fact check that could use from some community and tech help. Tools that help find where are misinformation or conversations about a particular misinformation is happening - subreddits, twitter, sharechat, whatsapp groups and find creative ways to plug fact check articles (without risking your safety and being blocked from the group that is)
Taking the statement ‘the impact of fact-checking is very limited’, I want to ask what are some obvious things that are being tried to improve that.
One challenge that I have observed myself is that most people who spread some fake news usually don’t trust fact checking services (I don’t need to expand on it, I am sure Pratik receives a fair share of those comments in his feed everyday).
But there are still people who rely on fact checking for their everyday news consumption. So when we say we want to increase the ‘impact of fact-checking’, is it targeted at increasing this user-base (or making more people trust them) OR making it more useful for people who already rely on it?
I personally feel that not every solution will cater to both goals, and some solutions might have opposite impacts on the two goals.
Kruttika Nadig - question for Venkatesh: could you explain what you mean by AI techniques for writing fact checks, and how would these help?
Muzayun Mukhtar asks what are the reasons that fact checked information doesn’t reach people as much as the fake news itself?
Fake news travels faster because of emotional arousal and novelty. Sadly, fact checks have neither of these.
Ragamalika Karthikeyan suggests that government regulations can’t be the solution in this world.
From Prateek Waghare: how do fact checkers determine at what threshold a claim should be fact-checked? Sometimes information disorder with low engagement can inadvertently end up being amplified. Alternatively the perpertrator can play the victim card to get more traction or as Joan Donovan pointed out by using internet archive etc (arguably the latter has not reached India yet).
To Pratik’s point information disorder is both a supply and demand side problem.
Question by S Parab: Do you see the race for wider outreach of fact checking corresponding to individual behaviour change? Does media consumption hinder the intent or motivate it? Are initiatives like yours maybe waiting waiting for an overton window for people to accept fact checking results of what we share and consume?
Taking screenshots of false content creates an archive of problematic content even if it’s taken down by the original poster later. Is there a potential for misuse by bad actors spreading the problematic content using the debunking article?
Once on internet, its difficult to delete permanently. There are ways bad actors can and will get access to digital content once posted, even if it is deleted later. Screenshots of false content, does not promote the original post/content as it doesnt go into neat, same, unambiguous class/bucket by id. So, while not perfect solution, its lesser evil, while still serving as basis for setting the context.
Manu Raveendran asks is there an annual report or analytics possible on how many fake news was tackled by social media platforms, AltNews, and so on? Categorized by themes? This probably gives hope and positive index of how well we are fighting fake news.
Transparency reports put out by tech companies contain aggregate numbers of content they’ve taken down for different reasons (including spam and hate speech): https://transparency.facebook.com/community-standards-enforcement#spam
Tattle has also opened a database of content checked by IFCN certified fact checking sites: https://services.tattle.co.in/khoj/
A couple of researchers at MSR tried categorizing the Covid specific content in this dataset by themes: http://joyojeet.people.si.umich.edu/an-archive-of-covid-19-related-fake-news-in-india/
Two Question for Prateek And Venkatesh: Q1 On what metric you decide to fact check a particular piece of information, Eg: Is it based on popularity or some other metric?
Q2: what about regional languages, How often fake news is created in any regional language ?
Rajeef M K asks if you could you share the research paper link here please?
I presume Rajeev is referring to fake news travelling faster, deeper, broader- paper by Vosoughi et al. http://fsnagle.org/papers/vosoughi2018spread.pdf
I also referred to this paper which says that fact checks may not completely convince a person that fake news is fake
Login to leave a comment