Hacker News new | ask | show | jobs
by jobenjo 4384 days ago
I wrote this flow about 3 years, ago, and my face is still the Twitter teacher (see twitter.com/finkel). I haven't worked at Twitter for over a year.

A few things. At least when I wrote it, the very first step you'd see the friends who tried to connect with you, if you have any. I doubt that's changed, so this may be a special case.

After building this flow, I helped my team design multiple better/more modern flows, but all performed worse than my original when we measured for retention over time. It turns out it's much trickier than it looks to build a better flow, and it also hard to prove that it's better due to bots/spammers.

Even though this flow is far from perfect, I still take pride that my face has been shown to hundreds of millions of new users, and that it greatly outperformed its predecessor. But I just want to remind users here that it's easy to say: "This is crap, I could do better." When it fact, many other "better" things have been tried, and it's surprisingly harder and more nuanced than it looks.

With my knowledge now, I believe more of Twitter's energy should be spent improving the product, because the new user flow is much less important than the product people see when it's done.

3 comments

You're right. Intuition isn't the same as data.

However I can see where the author is coming from as now I am engaged this is how I use Twitter too. Anecdotally it seems correct that people that get the best out of Twitter are those that interact publicly and productively with non-celebrity strangers.

One line seems to hit the nail on the head:

  > On Twitter, you interact with people you don’t know, 
  > and the friendship doesn’t have to be mutual. This is a 
  > good thing, but can be scary at first.
This is what makes Twitter great but it depends on growing your confidence and interacting inside communities where this is a norm. It can be quite socially dangerous to mess this up: some communities have high walls and aggressive in-group tendencies, and some people do not yet have the social skill required for global online communities.

The other day I read this great quote by Nick Szabo on Usenet and I think it holds true for Twitter, too:

  "Those who have never tried electronic communication
   may not be aware of what a "social skill" really is.
   One social skill that must be learned, is that other
   people have points of view that are not only different, 
   but *threatening*, to your own. In turn, your opinions
   may be threatening to others. There is nothing wrong
   with this. Your beliefs need not be hidden behind a 
   facade, as happens with face-to-face conversation.
   Not everybody in the world is a bosom buddy, but you
   can still have a meaningful conversation with them.
   The person who cannot do this lacks in social skills."
Twitter needs to connect like people to like people and slowly nurture positive, open conversation. But there are certain things you can't expect people to do with just a nice UX.

Perhaps they should be influencing with PR and education as well as UX.

Wow! Nice to meet you, Twitter teacher! Crazy that it's still the same three years on; I know what you mean that it's easier to talk about building a better flow than actually doing it, but I do think that it's fundamentally focusing on the wrong thing (from the outside in particular).

Perhaps this flow performs better than we think, but what I find most interesting about it is that instead of focusing on what you're interested in learning about/following, it instead just helps the bigger Tweeters get bigger/more famous, which is kind of sad, especially with many numbers suggesting that the average user has less than 100 followers.

I totally agree in terms of improving the product - it'll be interesting to see what Twitter delivers over the next year. I do hope that the focus shifts away from what it is now and more onto the intrinsic value of conversation/engaging with new people that in my experience seems to keep Twitter users active/proves the value to them.

Likewise. It amazes me how long it's lasted.

I agree with your sentiment about famous people. It feels really generic. I'm sure there's something better. We tested lots of interest-based versions, but they also underperformed.

Twitter is becoming more of a consuming platform for most users, so getting followers for the average user is much less important than finding great content.

I think you have to think of the twitter experience as a progression where you start with people that you know/have heard of/can relate to, and over time you discover new interesting people in areas of your interest to connect with directly. So I feel like it makes sense to first suggest following a list of people you may recognize and want to hear more from, and from them get drawn into using the service more deeply. This applies to recognizable celebs / media / etc as well as friends. the company can do a lot of analysis to only suggest celebs, etc that do a good job of hooking users over time.

That said, having worked on some even earlier versions of this process (and with Ben too) it is a constant work in progress and set of changes needed as more and more people join the network. Today's users are later to twitter and probably don't want to just start and get drawn in following individuals but probably want content quickly that is personalized to them and would be open to learn twitter from there. But following accounts may itself be too much work. Or the challenge is just that mobile Signup is so common now that no one gets the experience and starting point from the web. In any case, the biggest challenge and opportunity with onboarding is just to constantly be changing learning and trying to improve it.

Well said, Josh. It's about constant iteration and improvement.
The article sums up my experience well. I've had a Twitter account for years. I never use it. I have no idea how to use it. If I'm really bored I might take a look at Facebook, but never Twitter. I don't 'get' Twitter, but I want to (because I see many people I respect who apparently 'get' it), I just don't know how to.

My first thought was how you measure "better". You say "all performed worse than my original when we measured for retention over time". And this is not surprising - I doubt one flow works for all people. Did you ever try multiple flows to rescue stranded users like myself?

For example, whatever stats Twitter has on me right now shows that my engagement is zero. Would Twitter have better luck getting me to actively use their service if they tried re-engage me? Or do they think I'm a bot at this point? :)

I'm in the same boat. A lot of my friends love and use Twitter regularly, but I've never been able to get into it. I feel bad that I don't get it, because millions of people clearly find compelling use-cases for it, but my experience hasn't changed even after multiple attempts to figure out the missing piece.
Speaking of use-cases, I found this part of the article interesting:

> This is disappointing, as most of Twitter’s value is not tied up in following celebrities. I suspect that many users sign up for the first time, follow mostly celebrities and then give up, since they never get much interaction.

It's funny because that's precisely how I use Twitter. I've been using it daily (hourly sometimes) for more than three years, and I have less than 150 tweets. But I follow a number of interesting people who post often, like webcomic artists I like, some indie game developers, some AAA game developers, some celebrities like deGrass Tyson and Alton Brown.

I follow a couple friends on there, but mostly I find them to be annoying noise. I don't really give a shit about baby photos or what beer you had at lunch today.

People really use Twitter to have conversations? That blows me away. I have trouble making two sentences fit into 140 characters, much less an entire conversation. What kind of conversation can you have in that space?

Seems that you use twitter as "rss with human faces attached"