Developers hoping to access the T-Mobile store won't know for sure what their fee split will be with T-Mobile for offering the application on the phone (it depends, according to the article, on applications' network demands). Will T-Mobile host developers' free apps?
Quoting one gushing developer, the Washington Post suggests the T-Mobile store is the real story of the year:
"The App store was a big deal, but that's one phone. This is an entire carrier." In other words, we are talking about T-Mobile's 31.5 million subscribers today vs. the 10 million iPhones Apple expects to sell by year-endT-Mobile, unlike Apple, has several phone platforms to contend with. Developers interested in targeting T-Mobile's 31.5 million subscribers (is that U.S., or worldwide?) would need to deliver apps capable of running on the entire range of handsets T-Mobile allows customers to operate on its network. Presumably, some of the phones on T-Mobile's network don't have the computing power to do much in the way of interesting applications. Among those that do have the processing power, developers need to figure out which ones have a little finger-joystick, which ones can take a stylus tap or a touch, and which ones have always-available physical keys (and how those keys are labeled, and how they line up with the screen so on-screen indicators can suggest the right buttons for various functions). Without knowing something about the physical layout of the input devices, it could be hard to know what to draw on the screen, or which way the screen faces when the input devices are within reach.
via Washington Post (emphasis added)
This, of course, assumes the phone in question is one running an operating system for which the would-be developer has a developer kit. Developing without the dev kit is kinda a chore. Do we know the breakdown among T-Mobile phones by software platform? Are there many RIMM devices, or are they mostly Windows Mobile? Will T-Mobile support Android? Do Symbian devices get used on the network? Of the Windows Mobile devices, how many are actually going to support an application of the sort envisioned by our hypothetical developer?
The diversity of the hardware in the mobile space has been commented on at Daring Fireball as a reason developers might have a hard time providing a consistent experience to users even within a single software platform. The possibility that this problem hampers the growth of mobile devices as software platforms may give rise to a Microsoft cell phone, but given the possibly declining share the Microsoft music player has arguably attained, this might not be a big deal.
Apple's announcement that it's averaged over a million dollars a day in application sales (despite most apps being offered without any fee at all) offers some reinforcement that Apple might make significant revenue from the store. What is missing from this announcement is that the entire ecosystem around the App Store is gaining value, whether Apple makes a cent on the store or not. If the ticket into the ecosystem is iPhone purchases, Apple need not earn any net from the store to succeed from its existence. The fact that developers see they are keeping 70% of the store's revenue and that the revenue is enormous and growing will tend to bring developers to the platform. After all, that's where the app-buying customers are, and where the easiest store to use installs its software.
The platform being delivered in the form of iPhone hardware and software is becoming as potentially important to Apple as Macs: it's a platform that can reach a much larger audience (people who can afford and use phones on a daily basis being a larger population than that which can afford and will need computers), and it offers Apple a way to sell software (and upgrades) and services (and service support) to organizations and individuals worldwide.
The biggest problem facing Apple may be Apple's readiness to offer commercial services to a demanding world.
The combination of a sought-after hardware device, a software platform for which developers gush and plan development, and the ability to package all one's localizations in a single application bundle and to leverage user participation to create new localizations may yield Apple a platform trifecta: easy development, easy localization, and consumer demand sufficient to promote content. Even if individual applications can be replaced (ePocrates exists for numerous platforms, and is a free download), the experience -- the convenience, the integration of parts -- may be so good users are loathe to leave. As the platform grows, the possibility of lock-in due to unique applications certainly arises, but I don't think that this is the lure of the iPhone.
In these early days of the App Store, the iPhone sells without much "platform" value enhancement or lock-in from third-party apps; nevertheless those sales haven't been bad. Down the road, I expect better still.
 Here, "easiest to use" is intended to include the developer's ease in getting a free SDK, not having to worry whether his customers' hardware will be compatible, and so on. The $99 one-time membership to get the digital certificate needed to deploy on strangers' hardware may be the lowest barrier to entry of any mobile platform. When Android deploys, the mystery of figuring out what kind of hardware the developer is trying to program for will probably be a bigger problem, even if you neglect the cost of the web hosting needed to make the application available to (hopefully) curious seekers.