Last week Twitter announced a new tool to stem online harassment: the ability to download lists of blocked users in Excel-readable CSV data format from one account, and then upload them for use in another account. "[We] recognize that some users---those who experience high volumes of unwanted interactions on Twitter---need more sophisticated tools," writes Twitter user safety engineer Xiaoyun Zhang in the release announcement.
There's a problem here, though: While the intentions are good, the implementation is bafflingly underwhelming, especially coming from a marquee technology company. In fact, even calling it "shared" blocking at all is a misnomer, because there is no sharing mechanism to speak of. Rather, the CSV file is provided, speeding up the volume at which the blocks can be instituted, but the sharing of that file has to happen through some other third-party service---Dropbox, Pastebin, email attachments, or even handing off physical USB drives. Twitter specializes in the dissemination of text, extensive embedded metadata notwithstanding, but in this instance it has specifically avoided performing that role---even shirked a responsibility, one might argue, in continuing to place the onus for performing the actual updates on the abused.
It’s difficult to overstate how technically simple a CSV export is; given a list of items, any seasoned programmer can bang one out in a matter of minutes, perhaps an afternoon at most, if there are a few monkey wrenches hidden elsewhere in the codebase. So what we have here is not a sharing feature, but instead purely an import/export function---the sort you might also use elsewhere for backing up things you want, as opposed to those you don't. In fact, decades ago, before always-accessible online directories of contact information could be taken for granted, early chat programs like AOL Instant Messenger used to force you to do the same thing to transfer your contact lists between different physical computers; Twitter is doing exactly this, but between user accounts.
This is especially mystifying since the Twitter API already supports blocking from external apps and scripts---which is to say, blocking users is available as a controllable action to any developer who wants to write logic which triggers it. Other people have already designed third-party blocking tools which are far more robust than Twitter's new built-in feature, but native inclusion matters a lot.
Jacob Hoffman-Andrews, a senior staff technologist of the Electronic Frontier Foundation, released the Block Together Twitter app last summer, which he still sees as the superior option. But he’s optimistic about certain aspects of Twitter’s version:
Philosophically, what's still missing from Twitter's new abuse prevention feature is an understanding that "social activity" isn't an inherently positive currency on the Internet any more than it is so in real life---"social" is a descriptive modifier, not a qualitative one. It’s easy to forget this since the major social networks are both popular among users and also influential since they route tremendous amounts of the Internet's traffic. Realigning the value of "activity" and "engagement" might be difficult within the ranks of a company in the business of monetizing them. Data is much easier to study than the absence of data.
Even in the direct actions it is taking against online harassment and abuse, Twitter still privileges user activity over user safety, and has not yet elevated abuse reports to a first-class member of its data set. If this changes, the first signal will be the launch of a more efficient user interface that doesn’t immediately require taking block list data outside Twitter in order to utilize it. Twitter is literally exporting the problem instead of solving it.
Even beyond the issue of Twitter's proposed solution being on par with AOL's contact list management circa 1997, the mere creation of a list doesn't give due respect to the phenomenal complexity of online hate and abuse. The list is black and white – an account name is either present or absent, the CSV list item is either imported or it isn't. But there is no single standard for harassment online; it's not a simple toggle switch, so the tools built around it can't be as simple as binary filters. If you pay close enough attention, as by now Twitter certainly ought to be, most forms of idiocy turn out to have subtle degrees of nuance. New technology projects often benefit from being built in similarly iterative development stages, but not this one.
