The Great App.net Mistake | Hacker News Reader

The Great App.net Mistake

wired.com

120 pointssylvinus65 comments

Mat Honan deserves credit for one of the clearest, most concise descriptions of what App.net is (or intends to be):

> Imagine this. You sign up for Vine, and build up a robust friend network and library of videos. But then you try out Instagram’s new video sharing, and decide you like its editing features a lot better.

> Normally, this would mean starting over, with no friends and no files. But let’s say that both of them were just applications that ran on top of App.net. Instead of starting over, when you fired up Instagram for the first time, your friends and videos would be there waiting for you. That’s App.net. Or at least that’s what it wants to be.

Up until I read those two paragraphs, I couldn't shake the notion that it was really a Twitter clone that didn't want to be pigeonholed. Now I get it.

Huh? I'm actually more confused.

Doesn't Facebook-connect (and FB apps in general) effectively give you the ability to buy/rent their network effect already? Thus eliminating the "no friends and no files" notion?

Yes, but neither Twitter and Facebook are in the "social provider" business - they're in the ad business. This is why they've had so much friction with their 3rd-party developer ecosystem. Dalton wants App.net to be a social utility - providing ownership of social data to the user for a price. There are no ads, so there's no motivation to screw over developers.

That's the idea, at least.

Facebook is picky about which app is allowed to use your data. Twitter, for example, is not allowed to use it anymore. So when you develop an app, just relying on facebook could lead to a situation where you suddenly don't have access to your friends anymore (from within that app). Most apps also have their own user databases and just use facebook data for import. With app.net that's the one point where your user data would be saved.

Wouldn't those apps just be App.net clients? If you're limited only to what types of media App.net provides, how can you make an innovative app?

Because you can have your own backend that does anything you want, and just uses App.net for the social graph.

At that point wouldn't you be better off using a bigger social graph?

Yes - if you could get access to "a bigger social graph" without risking the owners of those bigger graphs shutting down or limiting your api keys, competing with you directly after you've proved/developed a market, or any of the other "dirty tricks" Twitter/FaceBook et al. have pulled on developers.

(Note: app.net might pull those dirty tricks in the future, but at least they've got a business model that doesn't _require_ them too)

Twitter only allows you access to 100k users or so (tokens), and your business is at risk of being turned off. or App.net providers you access to 150k+ users growing user base. Plus an a free sign up, access to hosting for file storage, and permissive messaging.

IMHO that is only true if you want to be a Twitter clone or otherwise be in the same business as Twitter.

Not if you want to leverage an already existing social graph, so that people have to sign up for one service less.

You can have your own backend that does anything you want, and is completely opaque and invisible to that oh so valued social graph, because /their/ clients don't support your magic thing.

The fact that you said "clients" instead of "apps" shows that you still don't understand the vision. App.net provides functionality that you can safely build an app on, and which is fundamentally more useful than the sandy foundations of Facebook or Twitter. That functionality is something which is non-trivial and which has potential network effects. You can use that functionality, you can extend it to something private to your own app, or you can mix and match with all the compatibility compromises that would entail. You can do all this safely without worrying that the rug will be pulled out from under you just when you figure out a magic formula that works.

In short, App.net is something with the potential to build a business on, whereas anyone trying to build a serious business (as opposed to just doing something cool) on Twitter is an outright fool at this point.

apps are clients.

what on earth makes you think they are not?

I understand the vision. But I also understand, as someone who has used app.net, in actual fact, simply porting your social graph into a new app does you no good. You still have to convince all your friends to also get that same app before they can see what you post with it. The vision only works with basic media, like text, and photos, posted in a very straightforward twitter-like way.

I'll give you a specific example: Group chat with photos. Yep, there's an app.net app that does that, and now, you can't see the photos unless you have that app. so what's the point? what's in the vision for this situation?

What group chat app is that? If you're talking about photos in patter chat, there's an established annotation for it that most clients can read and use.

Whisper, patter, Netbot doesn't support it. I don't know of any application other than whisper/patter that does.

My basic point is, the app.net vision is fundamentally flawed. It's not just a matter of positioning/marketing. It's the same fundamental flaw that google wave had. The vision is that app.net is a protocol, but you can't market protocols to people. you can only market apps. or more specifically, you can only market specific solutions to specific problems, or some fundamentally human vision (rather than a technological one)

With the app.net vision, any app you get that happens to use app.net as a back end will be subtly and frustratingly broken in some way because there's this feature, or that feature that your friend used that you can't see, because you have the wrong app. There's no commonality, there's no ultimate app that does all the things. With twitter, at least you get a link, something everyone understands, to a webpage, read in the ultimate universal client-a web browser. On app.net you just get mysterious silence. and "oh, that didn't work for you? huh..."

app.net is the linux of social media. Nerds love it but mortals will never get on board.

In an ideal world services would give you YOUR data, instead of locking you in, allowing you to easily export all your data -- ideally in an open format -- that could be imported to other services.

It's much worse than that. Exporting and importing does not preserve social connections. You need either federation or a benevolent monopoly.

App.net does provide an account export. But more important they've embraced open standards, see: http://blog.app.net/2013/08/07/response-to-brennan-novak-par...

This type of interoperability combined with sections of their API that are public (tokenless) provide a way to consume App.net content without needing an account unless you want to participate.

Then there are 3rd party apps and additions that bring even more open interoperability.

For development, there is Dev-Lite, so if you want to play around with the API, you don't need to spend $100/year until you're ready to join their Developer Incentive Program. And most Developers that have built anything for app.net have made their $100/year investment.

And App.net's annotation are another great example of creating new open standard of interoperability. These attachments can be added to any App.net API primitive to provide additional structured data to enhance interoperability between developer applications. Here's a link to their document repository for this: https://github.com/appdotnet/object-metadata

If App.net it's trying to be a "social backend"... if that IS the case, then their work it's harder than just being a "Twitter alternative". But I'm really hoping they manage to make it!

I had it all wrong. App.net is the new Ning. Or, I still have it all wrong.

Wow. I recall Ning. So many "social networks" on Ning lasted what, an hour? And to think, my boss was secretly orgasiming at the notion of the company bundling private social networks on the cheap through Ning. D'oh.

Except even this doesn't make sense. Presumably Vine handles the videos on its own backend, and you only get your social graph when you move over to Instagram.

Unless you're going to provide free infrastructure for every component (not just the social graph) there will be an element of lock-in to virtually any social product.

App.net's API does providing more than a microblogging API. They have a file hosting (each user has their own quota), permissive messaging (point to point or group), and a search API so you can easily find anything on their system. They're planning to add more useful APIs, such as billing.

Seems like bad business strategy to build on top of those if it means it's easy to steal away your userbase. Sounds fine for mid-size apps who can't justify rolling their own though.

If your only competitive advantage is user lock-in, then you already have a problem. Either you need to get your userbase to grow big extremely fast, or you've already lost.

And if you grow the user-base fast you've won, right? Lock-in and network effects matter enormously. Saying you can't lock-in the social graph neither guarantees a company won't lock-in elsewhere nor guarantees this is an appealing platform on which to develop.

This isn't a morality tale, I'm just pointing out what kind of decisions would be made around App.net and that the Vine/Instagram example doesn't really make sense.

From the beginning I've quietly predicted that if App.net is ever going to succeed, they will eventually end up going with a more traditional business model where they charge developers for the capacity their applications use (in my mind, they would end up resembling Firebase).

Dalton will write up some charismatic blog post about why this is the right move, and how it was really part of the plan all along. Some people will eat it up; others will call bullshit; and the world will keep turning.

Then 4 years later, he'll write a blog post on why selling the company to Twitter/Google/Facebook will be the best move. And how burnt out he is, and how he's leaving to pursue "other dreams". And people will congratulate him.

Why does this sound so familiar?

Good point. In the last App.net thread I said they'll probably need a killer app to get people to sign up. But if people are mostly in it for one app, why should that app settle for a slice of referral revenue? More likely the killer app would prefer to own the customer billing relationship and send a slice of revenue to App.net.

I don't see why we should criticize them too much if they change business models, though. If this was so easy that you could get it right on the first try, it would have been done already.

I'll give App.net one thing:

They get more press on HN than any other irrelevant company in the world.

They have funding. Journalists like free lunches.

I think it's more that they have lower odds of success (at their current mission) than most startups, but if they're successful, a bigger impact. And fundamentally it's meta, so more interesting than any given startup -- there's more of a philosophical argument behind it than just a better designed chat client or whatever.

Well, I think they captured some imaginations, to the point where they were told it's impossible. So it's an interesting experiment. Especially since they've been success and on-target to deliver most of their promises so far.

It's a huge risk, that not many investors would take. The financial returns are questionable, but what it could do for the Internet as a platform is disrupt the current mega social providers and IMHO I think that's worth trying.

In other words, normal people that use Twitter or Instagram still have no clue what App.net is and the early adopters of App.net will eventually get tired of it.

Early adopters aren't tired yet because App.net's continually releasing interesting updates to the platform. And there's even a great example of 3rd party developers working together on their system to extend and support one 3rd party developer's chat system (patter). So each developer's project help retainer interest as well.

So as long as App.net keeps releasing new APIs and developers stay interested in the platform. I don't think early adopters will get tired.

I get that App.net is supposed about owning your own data, and sure there's a brief mention of PRISM at the top. But why I would want to write an app for it in the era of PRISM? How can I trust this notion of a "private cloud?"

What I'm looking for now is more decentralization and encryption than a walled garden.

I read the whole article and never discovered the mistake. Did I miss something? It sounds like app.net is doing fine.

"Alpha was the first product out of the gate, launched in reaction to something Twitter did. Subsequently, much development was in building Twitter-like things. So its little wonder that people associate App.net with, well, Twitter. And once people form an idea of what a product is, it becomes very hard to change that idea. "

To be more explicit, the mistake is on the part of the people with the wrong idea of what App.net is. At least that's how I took it.

Why is the idea that app.net is a paid version of twitter so pervasive? That's what I thought until today. Perhaps it's my fault for being wrong, but if that particular mistake is so widespread, perhaps app.net is also making a mistake by not addressing it harder.

    Why is the idea that app.net is a paid version
    of twitter so pervasive?

Because this is exactly what it was laid out to be when it was first announced.

http://techcrunch.com/2012/08/12/app-net-reaches-its-500k-fu...

It is like Twitter, except it's like Twitter circa 2008/2009. It is what (many of) the early Twitter users though Twitter was and what they wanted it to become. The problem is that

1) Most people weren't around during those times.

2) Those who were have either forgotten what it was like or have come to terms with what Twitter is now.

That is a shame. 90% of the criticisms I've seen of App.net were mocking it for being 'an open source twitter' and completely ignored its ambitions to encompass general social network features.

Or that it's not open source.

It's not like there aren't F/OSS Twitter clones... just look at StatusNet.

Or Pump.io, which (AFAICT) is a Facebook/Google+/Diaspora clone.

Are open standards not important? Just the source?

They have been starting to release the source code of the client applications they've built: https://github.com/appdotnet/

They wanted to build an open, paid version of the Twitter API/ecosystem. But the value of the Twitter API lay in the access to their 100 million+ users, and all the valuable social data generated by those 100 million+ users. No amount of "open infrastructure", no matter how clever can ever replace that. It's like buying 1000 data centers in every country, and building an entire system that can handle a billion concurrent users, and saying "Yeah, we can take on Facebook with this", then realize the hard part was getting users, not scaling the infrastructure.

Why do I have to read this comment every time App.net gets brought up? Is every conversation about App.net doomed to turn into a debate about whether it will succeed, and if not why?

In a discussion about an article written about whether App.net will succeed, that seems like a pretty logical thing to be talking about, no? It's not like he hijacked something completely unrelated and changed the topic.

I suppose that's fair, but I still find this line of thought tiresome. At least the article is arguing that App.net's problem is that it's too much like Twitter when it could be used for more, rather than the usual tired argument that Twitter's success is simply network effects, therefore App.net cannot succeed. After all, as a business, App.net is already a success; it doesn't need runaway growth to be sustained.

It's not open. It's closed, proprietary and the boys over at Andressen-Horowitz are going to want a return on their investment.

They've embraced open standards: http://blog.app.net/2013/08/07/response-to-brennan-novak-par...

And they have started releasing client source code: https://github.com/appdotnet/

a16z is already getting a return, since App.net has been profitable for a while. It's all gravy since they've made it free to sign up. The user base is about 10 times bigger than last year. It's true we don't know how many new users are paying but since that fixed costs are covered, it doesn't really matter.

It was a big mistake. Clinging on to the twitter model gave them something tangible to model after and improve on but the result of that is the API is largely for micro-blogging. They have some file storage now but the primitives for a broad web application building platform aren't there yet. So from reading the API from a developers point of view, it's obvious that it's for micro-blogging. It takes a lot of imagination and effort to repurpose their API into something new.

The "mistake" is at the end...

'There are no breakout apps. There is no WhatsApp or DrawSomething. App.net remains a plaything for developers...'

...it certainly took them long enough to get to it.

While I appreciate Dalton's insight into incentivizing good behavior by treating a social network as infrastructure rather than an advertising/media business, I can't help but feel he's simply asking people to trust App.net because this time it's really different. What would prevent App.net from pulling a Twitter? Or from PRISM'ing our data? App.net seems like an incrementalist approach to a problem that requires more of a revolution. I wish there was something else with even half the traction of App.net if only it provided a distributed social network with no central point of failure and a solid API to build from.

If people don't want their social network to be at the whims of a company, why not go all the way and adopt the Tent protocol? http://tent.io

I mean, I can see why from a developer's point of view. Those $30k a month are enticing. But why would users care?

Devil's advocate: Maybe that $30K can pay for good UX which users definitely care about.

Funny you should mention that. I feel that crowd-sourcing user interface development by given an open access to an API will lead to new experiments with UI/UX.

I believe when Twitter was more open, we saw an explosion of clients and UI/UX. A couple of these innovations we hadn't easily seen before.

One of the hopes is that the App.net platform continues this experiment.

Users care because it's easier to pay $5/mo, than to set up and maintain your own infrastructure. Especially if you don't have any experience with servers (Think Grandma, Mom, Dad, non-techies etc).

Tent works like email-- some users run their own servers, but most sign up for a service that handles it for them (like https://tent.is)

Say what you want about it, but to class it as a mistake is wrong in my opinion. Back when Twitter was getting a lot of bad press from developers, the founder saw a market niche and successfully got backing from lots of people in that niche ... Enough devs were willing to stump up $50 to get something like the Twitter API without limits (or the client side ads).

I wonder if a part of these site or services is how the name works in the language. I can "tweet" something I can "google" something, but "app" something?

Not saying its crucial or 100% important, but I do think the name and it's ability to work well in English has a lot to d owith success.

Maybe?

The name is not a problem. You can “tweet” and “google” something because tweeting and googling use products. App.net is not a product, it is a platform. Similarly, you can’t “facebook” something, you can only “post on Facebook”, but that’s okay because Facebook is a platform.

Reply on news.ycombinator.com