We have entered a new phase of the discussion on what to do about speech on platforms such as Twitter. On the plus side many more people are engaged. On the minus side the calls to treat Twitter as a traditional publisher are growing (incidentally based on a misunderstanding of what being a publisher implies).
Let me start by repeating something I have written before: many of Twitter’s problems are self inflicted. In particular, Twitter has messed up its checkmark system and Twitter has been woefully slow to add moderation tools. But we shouldn’t base decisions about what to do in the future on Twitter’s current state.
There are many related problems when it comes to speech on a large public platform and often they cut in different directions:
Censorship and suppression of speech
Direct harassment and threatening of individuals
Hate speech
Misinformation and manipulation
Comments that are offensive to someone based on their beliefs
Being trapped in a filter bubble
The idea that there could or should be a single central institution, let alone a commercial company, which as a benevolent dictator resolves all of these issues to everyone’s satisfaction is a complete non-starter. Yet that is essentially what Twitter is attempting at the moment and it is, unsurprisingly, failing badly.
Here is just one example of the kind of problems Twitter’s current approach runs into. David Pinsen just had his Twitter account temporarily suspended. I initially connected with David on Twitter years ago, we have met in person for an interview and stay in touch with over Twitter, comments on my blog and email. Here are the tweets cited in David’s suspension:
Now people might disagree about how useful these tweets are. Some people might even have a negative reaction, such as “why did he have to bring that up?” and those people should be free to mute or block David. But to have Twitter suspend him, even for some time, seems like too much centralized power exercising censorship, i.e. fails #1 above. The calls for Twitter to be a publisher would massively aggravate this problem. Even a large traditional publisher has a tiny number of writers compared to the millions of voices on Twitter.
What should be done? Well, my preferred go to answer is to shift more power to the network participants by requiring Twitter (and other scaled services) to have an API. That would allow endusers to programmatically create the best version of Twitter and would also make it easier to simultaneously use Twitter and new decentralized alternatives. But since we are likely years away from accomplishing that via new regulations or removing existing regulations, let me set out what could be done within the existing legal framework.
For starters, Twitter should fix its verification system by having it simply mean one thing: this account actually belongs to the person whose name appears on the account. Together with some simple other changes, such as requiring verified accounts to have the person’s (or organization’s) name as their name (not their handle) and not allowing verified accounts to change their name without re-verification this would go a long way to dealing with misinformation and manipulation. Twitter would also need to make it much easier to actually obtain verification than they did historically (e.g. DM Twitter an image of your driver’s license from the account you want to verify).
To understand the next set of my proposals it is important to differentiate between the existence of a tweet and its visibility. Consider two extreme book ends: a tweet that exists but you can only see it if you have the URL for the tweet; a tweet that is inserted into everyone’s timelines repeatedly so that eventually all users see it. Twitter as a centralized system controls this entire range of visibility.
Twitter should never delete a tweet, unless either the user chooses to delete it or Twitter is forced by law to delete it. The latter should be extremely rare because even if one country asks for deletion of a tweet, Twitter could choose to make that tweet inaccessible from that geography based on IP or other geolocation technology. As an aside, Twitter could make tweets editable if it kept around all prior versions for the tweet as well and linked to them from the current version (it could also limit how many characters one can edit per iteration of a tweet and/or how many times one can edit a tweet).
Crucially, Twitter should significantly expand the features that let individuals and groups manage the visibility for tweets for themselves. There are already useful features such as muting a conversation or blocking an individual. These could be expanded in ways that allow for delegation. For instance, users should be able to say that they want to subscribe to mute and block lists from other individuals, groups or organizations they trust. One example of this might be that I could choose to automatically block anyone who is blocked by more than x% of the people I follow (where I can choose x). Ideally these features could be implemented at the tweet/conversation level and not just the account level.
The goal here is to retain Twitter as a platform for expression and empower individuals and groups to more clearly shape how they experience Twitter (but without immediately spilling over into what others can see). Now if a tweet is blocked or muted by more and more people belonging to widely different groups (something Twitter can tell through network analysis), Twitter could also gradually dial down the visibility of that tweet, but it would be doing so on the basis of a lot of signal.
There is one problem that this approach does not solve and may in fact worsen, at least initially. And that’s #6 above – people living in a filter bubble. This problem is endemic to any system that let’s people shape directly or indirectly (via algorithms) what they want to see. As it turns out people generally prefer having their existing beliefs confirmed rather than challenged. Asking any one system to overcome this deeply engrained human bias is asking a lot. This will require us to work on much broader changes along the lines that I propose in my book World After Capital. Nonetheless, with a delegated muting approach as proposed above, Twitter would have more data than ever that could also be used to selectively raise the visibility of some tweets in what I have called an “Opposing View Reader.”
No solution here will be perfect and it will take many iterations to get better. But to give up on platforms and revert to the publisher model would be a huge mistake.
