After the outstanding success of SMS, it feels like it ought to have been the next logical step. After all, who would have predicted that a simple text messaging interface to support engineering and customer support would have become a major craze and profit centre? Surely an even greater pot of gold must lie beyond the horizon once you make those messages more colourful, noisy, and marketroid friendly?
And thus was MMS (Multimedia Messaging Service) born. But it hasn't become a smash hit. People aren't sending untold billions of pictures and video clips to each other. They just hand their picture-filled handset round to their friends in the pub instead. The pot of gold was no easier to find than the end of the rainbow. Why is that?
In a sense, MMS is the pure antithesis of the end-to-end principle. Firstly break every message up into the constituent parts. Second, force users to categorise every part of the message as to what it contains by giving it a MIME type. Then make the network fully aware of the types of valid messages, and only allow those messages to be channeled via the carrier's own messaging gateway. Et voila! You can precision charge the user for the supposed value of everything that traverses your network. A different price based on message type, time, size, and destination.
For example, the price list from lousy UK 3G carrier three [PDF]:
[QUOTE]
*Charges for calls, fax and messaging outside any inclusive allowance for
your price plan*
| Text message | All UK mobiles | 10p |
| Photo message | UK mobiles | 25p |
| Video message | UK mobiles | 50p |
These prices do not apply to international calls and messages, calls made and received when abroad, premium rate calls and messages, directory service calls, calls to non-geographic numbers and special numbers. See page 19 for details of these charges.
Each text/photo/video message can accommodate 160 characters. Some handsets allow for more, these will be divided and sent in numerous messages (depending upon length).
[ENDQUOTE]
Wow. A pricing guru's wet dream. So what went wrong?
MMS is broken in many ways. The easiest one is that there's already a universally accepted multimedia store-and-forward system called email. It just happens to be free, unencumbered by obscure standards and patents, and detached from the telco-controlled numbering scheme (with it's paid-for membership). Clearly and awful technology to be avoided at all costs.
MIME types are too limiting. It just isn't possible to really encode enough meta information into the message. What's the MIME type for an urgent picture message? A rude video? The smart network can never have enough smarts to encompass all the unforseen richness of the world.
The next way it is broken is that the sending and receiving of MMS messages is dominated by the client built into the handset. Even if APIs appear in J2ME, BREW, Symbian and every other handset OS, vendors have to aim to the lowest common denominator. Additionally, they are afraid of liability if their app gets stuck in an infinite loop of sending MMS messages at 50 cents a pop. Thus your moblog service by default wants you to send a plain vanilla multipart MMS message using the supplied handset client. MMS actively discourages the development of highly useable purpose-built client applications. The single network gateway for messages gets coupled to a single user interface for messaging. The only thing worse than a smart network with a smart edge is a smart network with a dumb and inflexible edge.
But those are both really small fry. The biggie is that it confuses content with communication. Content tends to be one-way -- unicast, multicast or broadcast. Communication is interactive -- two way, many round trips, many actors, multisynchronous. But if you're charging a first-class fare for every one-way trip, the communication becomes unaffordable. And the richer the message, and the more valuable and compelling the user need, the more likely you are to need to engage in communication, not just content distribution.
The complete absence of messaging types in SMS ("just a string of characters, ma'am") meant that users layered their own communications protocols on top. "QUOTE IBM" gets "IBM 84.28 Mkt Close 7/16/04". "R U OK MUM?" gets "GR8!". The price was low enough that nobody cared too much. You could re-purpose it to meet most needs. If you'll excuse the computer science pun, every message is untyped. Thousands of de-facto applications have been created on the semi-dumb transport mechanism. The triumph of SMS was that it did a very simple thing very well.
Any richer media application typically involves a series of requests and responses. Post weblog entry ... entry posted OK ... view weblog entry ... edit weblog entry to correct spelling ... repost weblog entry ... entry posted OK. The pleasure of sending a simple picture message comes as much from seeing or hearing the reaction of the recipient as the act of sending itself. Searching for an apartment or a used car while out on the road isn't a fire-and-forget affair.
The two sides of the equation -- request and response -- don't necessarily have the same value. A request from me to E*Trade for a stock quote has no value to me. Only the response has value to me. A query from my picture messaging store asking if I really meant to delete every picture has more value than the response I send. So there's already a double-charging element forcing each transit via a toll-gated messaging system.
Even worse, multiple requests and responses in a conversational interaction don't scale linearly in value. If I want a graph of the IBM stock price, for the last year, expressed in pounds sterling, the price of the request shouldn't depend on whether I ask that as one question or through three steps (symbol, time and currency). And the cost shouldn't depend on whether the response is sent as a single yearly chart or divided quarterly.
Let's illustrate with a real example. A true compelling human need that might have real money attached to its fulfillment. One that demands sight, sound, color and text. How about getting laid on a Friday night? That'll do nicely. You want to browse a dating site for your perfect weekend rendezvous. You've got half an hour to kill on the train home from work. Now, do you see yourself prefering a dedicated application that lets you flip quickly through photos and descriptions, and flirt with others online (since there's integrated presence)? Maybe engage in some anonymous voice messaging? Or do want to use MMS and be charged for every time you hit the SEND button? It's no competition.
So MMS is, from the user's perspective, inferior in almost every way to a good old TCP/IP connection from a made-for-the-job handset application to an arbitrary Internet server. MMS lacks flexibility, straightjackets you into a single user interface regardless of the application, and costs way too much to do anything of value with it. Technically broken, philosophically broken and economically broken. It deserves to fail.
UPDATE: More here.
Posted by Martin Geddes at 11:13 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/270
Listed below are links to weblogs that reference Misconceived messaging service:
»
Misconceived messaging service from Firewall
MMS or as Telepocalypse thinks it should be called; Misconceived messaging service is again taking a whacking. There aren’t too many people that think MMS is a good thing. I've commented on it in the past. Tom Hume also quoted
[Read more]