578 karma · joined October 26, 2012
With a different algorithm Twitter could be a remarkably different experience. The current one has a strong bias toward engagement & velocity. It amplifies anger, outrage, and bad faith arguments over less "engaging" content.
If Twitter let off the engagement pedal a bit and re-factored the algorithm, our political climate might be less hostile overnight.
These companies have long claimed their algorithms are politically unbiased. That's not true. While I don't believe they lean left or right, they certainly have a strong bias towards extremism. Extremism keeps you engaged, which maximizes the time you spend in their apps.
Imagine if Twitter instead decided to bias their algorithm towards positivity
I'm fairly biased here, but I'd argue that product vision is often best executed in the hands of product designers rather than PMs. Having a design leader who takes the product or feature from high level goals, to flows & interactions, and finally down to the pixels often leads to a less-compromised end result. I usually find that singular vision beats design-by-committee.
Of course, this rarely happens due to company politics, which PMs tend to excel at versus designers.
They're trying to maximize user engagement/addiction and it's working. But what cost is their algorithm imposing on society? We're seeing more outrage in politics, more clickbait in journalism, and more privacy violations like this. Engagement-obsessed algorithms have created massive externalities which we all pay for.
The tech giants' business models strongly incentivize maximizing the time a user spends within their app. To do that, their algorithms try to find the most engaging content to show you.
The most engaging political content is often hyper-partisan, outrage-driven, and unconcerned with truth. These algorithms encourage bad faith participation by all players: media outlets, individuals, and politicians. Political ideas are often shared not out of deep seated belief, but in hopes of distribution.
Democracy seems predicated on the idea that most participants will engage in good faith. Whether we like it or not, our civic discourse has moved over to platforms that seem to incentivize bad faith participation. I think that's cause for alarm.
It will be interesting to look at the effective CPM rates of various candidates on Twitter using this data. I'm hoping they don't just show how many people were shown the Tweet as an ad, but also those who saw it from organic distribution via retweet.
I think you'll find that outrageous, incendiary politicians pay a lot less per view than reasonable ones.
Screen resolutions and pixel density have increased dramatically since HN was designed. It could at least use a fresh coat of paint :)
For example, digital property rights might be best preserved via decentralized tech. But perhaps you might take your digital property and use it to interact with centralized services. Imagine a video game where the player truly owns the in-game items they use as crypto tokens, but the game itself runs on centralized servers.
The same might be said of Facebook-like applications. Rather than imagining a future where a decentralized Facebook exists, perhaps it becomes more of a hybrid decentralized-centralized platform.
I think until recently the algorithm making the decision has had a peculiar bias. It's had a bias for "engagement" and the most engaging content is often the most outrageous / incendiary / clickbaity.
So if Facebook doesn't "choose what's good for you", a default decision was already made: they'll show you what's engaging. But I think it's important to appreciate that default is still biased. It's just biased in a strange way that could have negative political consequences.
I'm a big fan of decentralization in general, but I think it's worth pointing out where things could go wrong at scale.
Political campaigns are waged within the media landscape of the day. When Nixon sweated during the first televised debate, he lost the debate and the election. That signaled a transition from radio to TV as the dominant medium for political campaigns, and the types of candidates who won national office in the TV era were different from those from earlier eras. In that way the dominant medium of the day has a filtering & selection effect on who succeeds in a political race. Trump's election signaled another transition, this time to the era of social media. It's not partisan to worry about the type of candidates that transition will push to the fore.
Imagine a de-centralized Facebook with the same engagement-optimized feed. Maybe it runs over the Web, maybe it runs on something like Ethereum.
What if we built that, and wound up with the same set of problems? Except now there's no off button, nobody to petition, and government can't intervene.
Decentralization is not a silver bullet. And in this case I might worry that the medicine is worse than the disease.
As Jaron Lanier pointed out in his recent TED talk, with previous information technologies like books we figured out how to both charge for the information being distributed while providing access free of charge to those who could not afford it via public libraries. https://www.ted.com/talks/jaron_lanier_how_we_need_to_remake...
Furthermore, what Feld and others are describing here is an economic externality. You could both be right that many people would not and/or could not pay $5/month, yet still the negative social cost of a free product costs that user far more in the long run in the form of political unrest.
One obvious use case is collectible items for games. People spend thousands on games like Hearthstone or Clash Royale. Imagine if the cards you collected in such a game were ERC20 tokens. Instead of all that value being locked up within Blizzard or Supercell's servers, you could freely trade or sell your in-game items. I think that would encourage players to spend even more money on these types of games, and could open up huge new marketplaces for digital goods.
edit: and now back above $10k
Beyond Bitcoin, if crypto tokens begin to supplant public stocks (which I'm not saying they will), it would be incredibly foolish to cut off US citizens from those markets