Tag all content: mobicampbos
Handset secrets: from OEM to consumer
Dave Mitchell, ConnectedBits
5 years in mobile app dev
His company builds push-to-talk client, burned into ROM, on handsets for AT&T, etc.
Biz decision: avoid walled garden by putting yourself within the wall
BREW: aggregator handles the TBT for you, Verizon gets 50%, aggregator gets 50%
AT&T: too slow, can take 1 year from start of talks to live to consumers
Sprint: easiest, most innovative
Mobile web: not as good as rich apps
1 billion mobile phones sold last year, 700 million the year before
iPhone: great because all users go through iTunes
BREW: good, but hard to get through gatekeepers, and your mother is afraid to press the Shopping Cart button.
5 years ago, Dave didn’t see obvious path to get stuff on the market. Started out contracting for other app producers.
AT&T gets $10/month for the push to talk service. All the carrier cares about is monthly fees.
Wants consumers to have easy friction free way to obtain and use app’s
Like iTunes for the idepedent music artist model: put it out for sale; if it’s good, people will buy it.
Windows Mobile costs $2-3 per handset. Very expensive in the view of Asian manufacturers, who want $0 or $0.01, and nothing else
Android: no system for discovery of app’s. Not easy enough for your mom.
Finding app, getting it on your device, paying for it: too much friction.
Terminal: phone, device, handset
OEMs, component vendors: Chipset producers. E.g., TI, Seimens,
ODMs: original device manufacturers, produce the terminals, assembled from OEM parts. E.g., HTC.
Operator: the phone company, carrier. E.g., AT&T.
Operator spec’s what they want, issues RFP
ODMs bid on it
Operator and ODM negotiate on price and features. Maybe 10k handsets/month.
ODM produces phone, starting with early prototypes
Software developer adds app when handset is stable enough
3rd party testing, done by carrier’s contractor
Carrier internal testing
FCC, UL, battery testing, etc.
Bug fixes, retesting
Typical time span: 9 months
Repeat: bug fixes, 2nd release
AT&T 13340: document that specifies requirements for all phones
Same process for smart phones and feature phones
Carrier has complete control over phone’s feature set
Kodiak: push to talk server company
28 week test cycle for server hosted in AT&T’s NOC
iPhone changed the game: hook it up, and you instantly get 2 new features and 3 bug fixes
Phone company loses money when you call for support. Each call costs $20-$70. They lost money on you that month. Risk avoidance!
AT&T is very conservative because of the support costs.
Marcus (Yankee Group): interested in the impact of open access, especially given 700 MHz spectrum. Open access lobbied by Gogole. Verizon opposed. FCC required open devices, open applications. Anyone who bids in this spectrum must be open. More oriented around open access for devices to the network, not app’s running on the devices, e.g. any GSM device runs on any GSM network.
Category: like mix-in
Sort of looks like Lisp: [steveBalmer eat:@”Bacon-Wrapped Twinkies’ withPerson:”BillGates”]
Garbage collected, but not on iPhone
XCode, Simulator, Interface Builder (not yet available), Instruments
Must develop on a Mac
Buy a certificate for $99 to be able to deploy to and test on iPhone. Cert is keyed to your iPhone, not all iPhones.
CoreLocation framework: returns lat/long of phone. Doesn’t work in Simulator.
UIKit: buttons, etc.
Quartz: graphics, points, lines, rect’s, animation
HIG: Human Interface Guidelines
Well designed app’s
Leave things out
Sensible defaults and few pref’s
Integrate with the platform. Use the built-in stuff, like contacts and UI. Follow platform’s customs.
Simulator screen size vs actual screen size: simulator is larger
Fingers, not mice
Avoid keyboard input
Contrast: readability varies in different light conditions
Lots of fan discussion, especially around the keyboard
- Chatted with Matt Gross of ULocate about LBS (location based services) and ULocate's Where platform.
- Where is an app available on all carriers. It is a widget framework. Developers build widgets that run within Where.
POI: points of interest
TeleAtlas, NavTeq: the two companies that provides all digital maps for all app’s
BrandIcons from TeleAtlas
Displays, e.g., Shell logo at location of Shell gas stations on personal nav device display
Location based advertising: growing to $19.2 billion by 2011
POIs are preloaded on nav device, updated occasionally
Not trackable: user views it, but no one can track views
Advertising that’s not really advertising, like Google: you already qualified yourself by searching for a particular thing; it’s just a search result.
Possible future: Google wins 700 MHz spectrum auction, sets up cell coverage nationwide, gives away phones and phone service. How do people make money? Google would make money through paid ads.
Converted incremental sales: better than pay-per-click
Google uses TeleAtlas for API, NavTeq for consumer maps
Comment: on-deck doesn’t matter because soccer moms have never seen it—they’re just discovering SMS.
Pragmatic development for mobile
- My talk, with Dave Mitchell. I will publish notes soon.
Mobile advertising for developers
- My talk. I will publish notes soon.
- Very few people in the community are doing what we are doing at Nellymoser: Mobile 2.0, partnering with giant carriers and media companies, in-app advertising, preload app