After much deliberation and hand-wringing, the founders of App.net laid off their entire staff and put their servers on autopilot in an effort to keep the service afloat with its remaining pile of money. Pundits all over the web have weighed in on this decision with varying degrees of completeness, and I have no intention on saying what could have been or should have been. Instead, I am going to propose that the founders of App.net let go of various aspects of the service, granting a 3rd-party the responsibility of maintaining and expanding an operation.
This is the first in a series of posts outlining a possible future vision of App.net as an infrastructure, and the various tools built on top of this infrastructure.
Part I: Alpha Needs an Owner
Part I.I: Clarification on Alpha Ownership
Part II: Thinking About NiceRank
Alpha, the microblogging tool initially built to showcase some of what App.net could do, has been adrift for far too long. In the last year there have been no real feature updates and almost no activity from the employees of App.net. @dalton and @berg have been posting regularly but, of 10+ employees, they are the only two. This is a clear sign that the company didn’t have any love for the project and let it sit while working on … other stuff1, so why keep it?
I propose that Alpha, both the website and the communications API, be spun off to be its own service that operates on separate servers and is run by a separate group of individuals outside of App.net. The alpha.app.net sub-domain can be leased from the company, and the founders of the infrastructure can have input on the system, but it would be a separate entity ultimately under the control of other people. App.net did not seem to be very interested in the microblogging aspect of their service to begin with, so this should not be a big loss for them. And besides; the App.net servers will still be needed.
Please excuse me while I get up on my soapbox, but these are the changes that I would make to Alpha if I were the one to call the shots:
Alpha Goes Subscription-Based
Alpha would need to generate revenue in order to pay for the lease of the alpha.app.net subdomain, and it would need to pay contract developers2 to build out many of the changes that would need to take place in order for the microblogging tool to turn a small profit. Ideally the project would be run as a non-profit enterprise, and the operators of Alpha would need to remain accountable to the community.
How much would this cost? It really depends on negotiations with the founders of App.net as well as the costs of hosting the Alpha servers but, ideally, it would not exceed $4 per year.
That would mean a premium App.net account would be $36, and an Alpha subscription would bump up the total annual payment to $40. More on this in a moment.
Free Accounts Stay. Hello Client Restrictions. Bye-Bye 40-Account Limits.
App.net, the infrastructure, would continue to allow people to sign up for free, of course. Stopping this would be a foolhardy move as it would interfere with other future projects that may use App.net as its infrastructure. The genie cannot be put back in its bottle. The free accounts are here to stay. That said, a privately owned Alpha does not need to give them full reign on the network. Free accounts will be limited to a maximum of 50 link-free posts per day. Only the Alpha website client could send messages on the account’s behalf. Free accounts with a $4 subscription will be able to use links and use 3rd-party clients, but will still be limited to 50 posts per day.
Premium App.net account holders will be able to send 250 messages per day (with or without links — just remember that an image is considered a link). Just as with the free accounts, being a subscriber will enable support for 3rd-party clients.
Oh, and to heck with those 40-account follow limits. Follow as many people as you want. It’s your timeline, after all.
Did Somebody Ask For Lists? Want To Mute or Block Clients? You Got It!
This has been a long-requested feature. It will be one of the first items implemented in the new Alpha. As one would expect, you wouldn’t need to follow somebody to put them in a list. You can make a list of 100 media accounts3 and watch the stream go by without ever making a mess of your main timeline.
Want to ban IFTTT, Buffer, PourOver, 10Centuries, and any other non-human account from your timeline? You got it! Blocking clients, hashtags, keywords, and even account types will happen at the API level meaning that you can take your preferences to any client you want. There’s no limit.
These features will also be available to subscribers of Alpha.
I’ve been thinking long and hard about the true cost of running App.net and, to be perfectly honest, I have no clue how much it would cost. The servers are all hosted on Amazon’s EC2 service and, from my experience, this is an enormous waste of cash. ADN, the company, has undoubtedly paid more to Amazon for their under-performing, inconsistent, semi-reliable virtual servers than it would have cost to buy decent dedicated hardware from HP or Dell. Paying for bandwidth is also stupid.
These servers need to be moved offshore, and I think Japan would be a perfect location to host Alpha. Bandwith is often free in Japan. Virtual servers equivalent to Amazon’s c1.medium can be had for as little as $12 a month. Storage is dirt cheap. To make the move even easier to swallow, Japan is connected to North America, Asia, and Australia by multi-terabyte-per-second, undersea, fibre optic cables. The added latency would be negligible for 90% of the people who would use the service, and some would even see a speed improvement.
For the tiny amount of $4 a year from 3,500 core accounts4 would generate $14,000 a year in subscription fees. This should be sufficient to pay a small amount to App.net for the Alpha sub-domain, as well as provide enough for the servers and a little bit of contract work on the Alpha API. If more people subscribe, then more can be done to ensure the evolution of Alpha.
It’s such a tiny sum of money, but that’s why the non-profit model works for this situation.
Of course, if the true cost of running just Alpha later becomes known, this number may be revised.
Six Months Later
Within six months of these changes, spam would be pretty much eradicated from the network as people will not pay $4 per account to fill the global timeline with 250 links peddling wares nobody cares to see. Media companies will not see Alpha as a less-popular Twitter clone where it’s perfectly okay to shove 25 IFTTT-made posts of links to asinine, single-paragraph “articles" into Global every half hour. What’s more, people will be able to more easily use Global as a way to discover new people and encourage them to upgrade, either to a Premium App.net account, or to an Alpha subscription.
Perhaps at this point, it would then be possible to move on to stage two: fixing the Developer Incentive Program. (To Be Continued)
Update 1: @fam has informed me that App.net does not run on Amazon’s EC2, but dedicated hardware located on the west coast of the United States. This is good news, as it drastically reduces the cost of running the service.