This is really good. I have pondered that question of purpose a lot. My stab at it? "an API for ideas" - what do I mean by that? The joy of twitter for me was the combination of its ubiquity (everyone in the same chat room, or at least a huge number of them); its timeliness (when I see a massive cloud of smoke pouring from a city building, I don't look anywhere else than Twitter search for my first clue as to what's happening - there's almost certainly someone on the scene with info and pics); and its search. Again, real-time, and very easy to find people or topics. So for me it was always a vast database of sentiments, news & ideas, and careful interrogation and structure could so easily make useful things from that. Or even just silly things. I created the data structure for #uksnow on exactly that basis in 2009, and a few months later, started the tweetbike service during the tube strike. Write-ups here: https://paulclarke.com/honestlyreal/2009/02/a-flurry-of-uksnow/ and http://news.bbc.co.uk/1/mobile/talking_point/8093595.stm (somewhat degraded UI now, sadly)
I'd completely forgotten you were involved in UKSnow! I think it's a really fascinating example that touches on a couple of the themes here as well. It shows how if you give just a little bit of an extra hint to users, and incentivise 'well-formed contributions', it can lead to far more productive behaviour on a platform.
Whilst I am very new to Twitter, I did spend 5 years on Reddit which has the (unique?) selling point of having the replies upvoted (as well as the original post). This means that the best ones rise to the top. This disincentivises low effort replies (they sink to the bottom never to be read) and makes the conversation better to read by highlighting the best. This "crowd shaping" of the thread seems to be a key missing ingredient from Twitter.
Yes, that's a good point - one of my bugbears is the way that you can have some amazing tweet from a smart person, and then the first thing underneath it is some act of low-level vandalism - "your a dick!" - which then just wrecks the tone of the ensuing conversation. Crowd voting would be good, but my worry would be that the way people discover stuff on Reddit is different from Twitter - on Twitter it's easier for people with huge followings to set their supporters on a piece of content and aggressively downvote it. So I reckon for Reddit-style voting to work, you'd need to think about how people land on the content in the first place and maybe disable or change some other features to not sabotage it, if that makes sense.
As a compromise, perhaps do away with quote-retweets, including screenshot ones? Limit users to either joining the conversation (reply) or marking as worthy of attention (like). Also, if someone retweets something with a link in it and they have not clicked on the link themselves Twitter should quietly file it in the bin. More radically, you could score users on constructiveness of engagement. Users can set a threshold of constructiveness on what they see or who can reply / retweet. A users constructiveness score is affected by abuse, ratio of orig content etc. Basically, herd all the trolls into a corner of the platform the rest of us can ignore.
Hi, random guy here... Screenshot tweet is the final boss here, the one that is made just to make your followers angry about someone anonymous or not anonymous, and that removes any control from the creator. And I am seeing it everywhere in my timeline.
Yes, true - the screenshot RT is really pernicious. Even if you've blocked someone, they can still log out, view the tweet, screenshot and share. It's also problematic because so many screenshotted tweets are actually fakes. One technical solution might be for Twitter to process images on upload, detect ones that are screenshots of tweets, and filter those out (given the consistent format it shouldn't be hard to do.) But tbh you probably can't stop stuff like this completely. I see this a bit like lockdown measures in a pandemic - each component is a multiplier, so even if you can't stop every single vector of transmission, closing or reducing some of them has a multiplying damping effect. If screenshots were the only way to RT, it would still substantially reduce the problem, even if it didn't remove it completely.
I left this comment on Twitter (schoolboy error, I know...) so thought I'd post it here as well.
Interesting article. Not sure I agree with all these ideas but I think most important is to educate people to use all sorts of social media politely and that needs to be taught in school (and everywhere). So the idea of user-edited replies & being able to remove abuse seems good to me, although I'd still prefer replies to be more thoughtful and polite in the first place.
If we allow replies on Twitter to be edited or controlled, should Twitter introduce a 'dislike' option as well as 'like' so people can register disapproval without being abusive or rude? A net score on posts could be an interesting metric for anyone looking at a post. If there were some really awful posts that appal a large number of people then it would immediately highlight that fact to anyone coming to them fresh (although that could I suppose backfire if the awful people pile on the likes).
P.S. Copying and pasting my tweet here makes me realise how bad my typos were (I literally don't see the errors - it's a brain malfunction and illness, not laziness, honest. Would an edit function on Twitter be such a bad idea ? (even if it is just for a short period after posting rather than a permanent thing)
A few people have mentioned a 'dislike' button, and I think Facebook actually contemplated the idea for a while, before deciding not to go ahead with it. I recently started posting on Instagram, and one thing that struck me was how positive it feels there because you're a) creating something nice to share, and b) the responses are generally 'I like this'. I think one way of looking at friction on Twitter is that people share too much stuff they hate and not enough stuff they like, which makes the whole site seem negative. So not having a dislike button seems like a smart move, to help balance that. OTOH, maybe letting people click 'dislike' gets it out of their system and stops them posting more negative comments? It'd be an interesting thing to do user research on.
One of the joys of Twitter for me is finding new (quality) people to follow, through quote tweets and retweets, although that's also how negative voices are amplified too. Solving discoverability of the 'good' without amplifying the 'bad' is a challenge, although personal definitions of 'good' and 'bad' people on Twitter depend on personal preference.
Definitely. That's actually why I like the 'like' button more than the RT. Clicking either one allows your friends to discover content you've engaged with, but 'like' is a more positive reaction by default, whereas RTs and QTs are very often negative. So ditching RTs and keeping likes feels like a good way to rebalance the site toward people stuff they feel positive about.
I'm the exact opposite. I hate the @martin liked this tweet stuff pushed into my timeline. As I see it, the like button is essentially a message to the creator saying "well done" or at least "I've seen this", this is especially true in replies to tweets containing questions. The RT is explicitly about sharing content to your followers (for good or ill). The other problem is likes appearing in your timeline are at the mercy of the Twitter algorithm, rather than a deterministic process.
KA is absolutely right about the tension between discovery and amplification, and you're right to identify the negative nature of many RTs/QTs by certain sections of twitter. I just don't think trying to overlay this onto the like button is the right solution.
This is really good. I have pondered that question of purpose a lot. My stab at it? "an API for ideas" - what do I mean by that? The joy of twitter for me was the combination of its ubiquity (everyone in the same chat room, or at least a huge number of them); its timeliness (when I see a massive cloud of smoke pouring from a city building, I don't look anywhere else than Twitter search for my first clue as to what's happening - there's almost certainly someone on the scene with info and pics); and its search. Again, real-time, and very easy to find people or topics. So for me it was always a vast database of sentiments, news & ideas, and careful interrogation and structure could so easily make useful things from that. Or even just silly things. I created the data structure for #uksnow on exactly that basis in 2009, and a few months later, started the tweetbike service during the tube strike. Write-ups here: https://paulclarke.com/honestlyreal/2009/02/a-flurry-of-uksnow/ and http://news.bbc.co.uk/1/mobile/talking_point/8093595.stm (somewhat degraded UI now, sadly)
I'd completely forgotten you were involved in UKSnow! I think it's a really fascinating example that touches on a couple of the themes here as well. It shows how if you give just a little bit of an extra hint to users, and incentivise 'well-formed contributions', it can lead to far more productive behaviour on a platform.
Whilst I am very new to Twitter, I did spend 5 years on Reddit which has the (unique?) selling point of having the replies upvoted (as well as the original post). This means that the best ones rise to the top. This disincentivises low effort replies (they sink to the bottom never to be read) and makes the conversation better to read by highlighting the best. This "crowd shaping" of the thread seems to be a key missing ingredient from Twitter.
Yes, that's a good point - one of my bugbears is the way that you can have some amazing tweet from a smart person, and then the first thing underneath it is some act of low-level vandalism - "your a dick!" - which then just wrecks the tone of the ensuing conversation. Crowd voting would be good, but my worry would be that the way people discover stuff on Reddit is different from Twitter - on Twitter it's easier for people with huge followings to set their supporters on a piece of content and aggressively downvote it. So I reckon for Reddit-style voting to work, you'd need to think about how people land on the content in the first place and maybe disable or change some other features to not sabotage it, if that makes sense.
As a compromise, perhaps do away with quote-retweets, including screenshot ones? Limit users to either joining the conversation (reply) or marking as worthy of attention (like). Also, if someone retweets something with a link in it and they have not clicked on the link themselves Twitter should quietly file it in the bin. More radically, you could score users on constructiveness of engagement. Users can set a threshold of constructiveness on what they see or who can reply / retweet. A users constructiveness score is affected by abuse, ratio of orig content etc. Basically, herd all the trolls into a corner of the platform the rest of us can ignore.
Hi, random guy here... Screenshot tweet is the final boss here, the one that is made just to make your followers angry about someone anonymous or not anonymous, and that removes any control from the creator. And I am seeing it everywhere in my timeline.
Yes, true - the screenshot RT is really pernicious. Even if you've blocked someone, they can still log out, view the tweet, screenshot and share. It's also problematic because so many screenshotted tweets are actually fakes. One technical solution might be for Twitter to process images on upload, detect ones that are screenshots of tweets, and filter those out (given the consistent format it shouldn't be hard to do.) But tbh you probably can't stop stuff like this completely. I see this a bit like lockdown measures in a pandemic - each component is a multiplier, so even if you can't stop every single vector of transmission, closing or reducing some of them has a multiplying damping effect. If screenshots were the only way to RT, it would still substantially reduce the problem, even if it didn't remove it completely.
I left this comment on Twitter (schoolboy error, I know...) so thought I'd post it here as well.
Interesting article. Not sure I agree with all these ideas but I think most important is to educate people to use all sorts of social media politely and that needs to be taught in school (and everywhere). So the idea of user-edited replies & being able to remove abuse seems good to me, although I'd still prefer replies to be more thoughtful and polite in the first place.
If we allow replies on Twitter to be edited or controlled, should Twitter introduce a 'dislike' option as well as 'like' so people can register disapproval without being abusive or rude? A net score on posts could be an interesting metric for anyone looking at a post. If there were some really awful posts that appal a large number of people then it would immediately highlight that fact to anyone coming to them fresh (although that could I suppose backfire if the awful people pile on the likes).
P.S. Copying and pasting my tweet here makes me realise how bad my typos were (I literally don't see the errors - it's a brain malfunction and illness, not laziness, honest. Would an edit function on Twitter be such a bad idea ? (even if it is just for a short period after posting rather than a permanent thing)
A few people have mentioned a 'dislike' button, and I think Facebook actually contemplated the idea for a while, before deciding not to go ahead with it. I recently started posting on Instagram, and one thing that struck me was how positive it feels there because you're a) creating something nice to share, and b) the responses are generally 'I like this'. I think one way of looking at friction on Twitter is that people share too much stuff they hate and not enough stuff they like, which makes the whole site seem negative. So not having a dislike button seems like a smart move, to help balance that. OTOH, maybe letting people click 'dislike' gets it out of their system and stops them posting more negative comments? It'd be an interesting thing to do user research on.
One of the joys of Twitter for me is finding new (quality) people to follow, through quote tweets and retweets, although that's also how negative voices are amplified too. Solving discoverability of the 'good' without amplifying the 'bad' is a challenge, although personal definitions of 'good' and 'bad' people on Twitter depend on personal preference.
Definitely. That's actually why I like the 'like' button more than the RT. Clicking either one allows your friends to discover content you've engaged with, but 'like' is a more positive reaction by default, whereas RTs and QTs are very often negative. So ditching RTs and keeping likes feels like a good way to rebalance the site toward people stuff they feel positive about.
I'm the exact opposite. I hate the @martin liked this tweet stuff pushed into my timeline. As I see it, the like button is essentially a message to the creator saying "well done" or at least "I've seen this", this is especially true in replies to tweets containing questions. The RT is explicitly about sharing content to your followers (for good or ill). The other problem is likes appearing in your timeline are at the mercy of the Twitter algorithm, rather than a deterministic process.
KA is absolutely right about the tension between discovery and amplification, and you're right to identify the negative nature of many RTs/QTs by certain sections of twitter. I just don't think trying to overlay this onto the like button is the right solution.