Hacker News Reader

Comment by macca321 | Hacker News Reader

They built it.

They had an experimental project called 'annotations' where you could attach 1k of json to each tweet, like a DIY microformat.

I got onto the beta and created a prototype twitter client which you could attach mini 'apps' to tweets based on the payload type, e.g. you could tweet out a poll, or an inviation to play a game or a job advert or whatever, and you could attach your own app as a listener.

I think it could have been pretty amazing, basically A Message Bus For Everyone. If you build this yourself, you will run into the chicken and egg position, twitter could have pulled it off as they had the eyeballs and the developers.

Unfortunately they pulled the plug on it, around the time they started to close down the ecosystem.

View on HN

It's not surprising really, is being a centralized message bus really attractive when it just means that you'll be responsible for policing it?

If you run a central message bus off which people can produce their own applications, you'll find yourself having to deal with everything that ISPs and mail providers have to deal with. That means being asked to keep logs for years, dealing with DMCA takedowns, filtering the worst of the internet, etc.

And for what? As soon as anyone relying on it achieves scale they'll re-implement it themselves.

Besides, what you've described is the opposite of a protocol: an inner platform.

I disagree that anyone else could re-implement it. Once you have a critical mass of listeners, it would become the first port of call for any message you wanted to distribute.

As for policing, I think that would probably be where the money would be made - reputation/quality of the messages being posted.

Why would people pay additionally when they don't need to pay now (relative) to distribute/post messages?

Suppose twitter had become a giant json firehose, and you could listen to it and, for example, build a jobs listings website from the jobadvert microformatted tweets.

It would end up full of fake job spam, and the listeners might want quality filters. I can imagine that you might pay to validate your identity as not-a-spammer.

How would "Twitter" capture that value and not some third-party platform who and offers those filter services? The value and revenue isn't with Twitter then, so then how does Twitter pay bills? Or am I seeing this in a different light than you do?

As far as I can tell a simple freemium model would have worked. You charge for the firehose and anything high volume, but you give a generous free allowance to keep the grassroots ecosystem healthy. They could have played all the same pricing games that Facebook is now playing, except across a much wider range of channels.

I'm not saying it was a slam dunk, but a lot of talk about this back in 2007/2008 and there was a palpable excitement about Twitter apps. Within a few years the media-company mindset had taken grasp, the valuations and expectations exploded, and now we have boring old Twitter which is still a damn cool thing in its own right, but which is seen as a failure because of a combination of pedestrian vision and mismanaged expectations.

I'm not saying it would have been a good business decision for Twitter, but it would have been pretty neat for developers and users.

Wouldn't spammers (assuming they're spamming to make money, and therefore likely make more than $0 per spam) also just pay money to be validated in the same way?

Interesting thought. Legitimate posters could hopefully could out price spammers.

They could add several seconds of latency by default and sell high speed premium to trading bots. Humans don't care about a few seconds.

Humans definitely care about a few seconds, interesting strategy though.

Why would you want to build a service that distributes applications to mobile phones? There are many ways to make money.

What is the competitive differentiation and defensibility then in your mind?

I was thinking of a service that distributes applications to tweets.

"They had an experimental project called 'annotations' where you could attach 1k of json to each tweet, like a DIY microformat."

That's what we're doing with "Oh By"[1]. When you speak of a "message bus for everyone", that's our goal.

You are correct, however, that there is a chicken and egg problem - Oh By Codes are interesting if everyone knows what Oh By Codes are. Otherwise, people just wonder why a weird code is written in chalk on the sidewalk ...

[1] https://0x.co

Reply on news.ycombinator.com