last friday afternoon, Twitter released the source code of its recommendation algorithm to GitHub. Twitter says it’s “open sourcing” its algorithm, which I usually agree with. Recommending algorithms and open source code is a major focus of my work as a researcher and advocate for corporate responsibility in the tech industry. My research has demonstrated why and how companies like YouTube are more transparent about the inner workings of their recommendation algorithms — and I’ve campaigned to force them to do so. The nonprofit Mozilla, where I am a senior fellow, is known for open-sourcing the Netscape browser code in 1998 and inviting contributions from developer communities around the world, and it has been pushing for an open Internet ever since. So why am I not impressed or excited by Musk’s decision?
If anything, Twitter’s so-called “open source” is a clever diversionary recent move leave from transparency. Just a few weeks ago, Twitter quietly announced that it would be shutting down the free version of its API, a tool that researchers around the world have relied on for years to conduct research on harmful content, disinformation, public health, election monitoring, political behavior, and more. Replaced by it The tool will now cost researchers and developers between $42,000 and $210,000 a month to use it. Twitter’s move drew the attention of lawmakers and civil society organizations, including the Independent Technology Research Alliance, of which I sit on the board, who condemned Twitter’s decision.
Ironically, many of the questions that people were asking when they analyzed the source code over the weekend could actually be tested with tools that Twitter is disabling. For example, the researchers speculate that the “UkraineCrisisTopic” parameter found in Twitter’s source code was a signal to the algorithm to downgrade tweets about the invasion of Ukraine. Using Twitter’s API, the researchers could retrieve tweets related to the Ukraine intrusion and analyze their engagement to determine whether the algorithm amplified or de-amplified them. These tools allow the public to independently confirm or refute blocks of information provided by the source code. Without them, we’re subject to the truth of what Twitter tells us.
The Twitter stunt is just the latest example of the tech industry’s laundering of transparency. In 2020, TikTok also used the term “source code” to dazzle regulators in the US and Europe, who demanded more transparency about how the platform operates. It was the first platform to announce the opening of a physical “transparency centre,” said to be designed “to allow experts to examine and verify TikTok’s practices.” In 2021, I took a virtual tour of the center, which was nothing more than a Powerpoint presentation by the TikTok policy folks explaining how the app works and reviewing their already public content moderation policies. Three years on, the hubs are still closed to the public (TikTok’s website cites the pandemic as a reason), and TikTok hasn’t released any source code.
If Musk really wanted to bring accountability to Twitter’s algorithm, he could have done it Can be studied also transparentFor example, he could have created tools to simulate the output of an algorithmic system based on a series of inputs. This will allow researchers to conduct controlled experiments to test how recommender systems rank real content. These tools should be available to researchers working in the public interest (who, of course, can demonstrate how their methods respect people’s privacy) at little or no cost.
There’s good news on this front: Europe’s Digital Services Act, which will come into force for the very largest online platforms as early as this summer, will force platforms to undergo third-party audits of their algorithms to ensure they’re not at risk of harming people. The kind of data required for such an audit goes far beyond what Twitter, TikTok, or any other platform currently provides.
Publishing the source code was a bold and hasty move for which Twitter itself seemed unprepared: the GitHub repository has been updated at least twice since its release to remove embarrassing parts of the code that may never be made public. While the source code reveals the underlying logic of the algorithmic system, it tells us little about how the system will perform in real time on real tweets. Elon Musk’s decision makes it impossible to judge what is happening on the platform now, or what will happen next.
wired opinion Publish articles from external contributors representing a broad range of viewpoints.read more opinions hereand review our submission guidelines here. Submit column [email protected].
