Apple has delayed Leopard until October and, unusually, told everybody exactly why: the engineers who would have finished it have been moved onto the phone. I cannot remember the last time a company of this size admitted a trade-off that plainly.
Normally a delay comes wrapped in language about quality and polish and taking the time to get it right. What Apple has published instead is close to a resource allocation memo. There was one pool of people capable of doing this work, there were two products that needed them, one of those products has an immovable public date, and the Mac lost.
I find the honesty admirable and the decision unsettling, and I want to separate those two reactions.
The decision is defensible on its own terms. A June ship date announced from a stage is not a real constraint; Apple could have moved it quietly. The phone’s date is real, because carriers have been briefed, retail has been arranged, and a company cannot delay a product it has spent five months telling the world is arriving. Faced with slipping the thing that can slip, Apple slipped it. Any competent operator would.
What unsettles me is what it reveals about the size of the bet. This is a company whose entire business, until quite recently, was the Macintosh. It has now delayed the first significant update to that platform in years, publicly, in order to finish a product in a category it has never shipped in, against competitors who have been doing it for a decade. Nobody outside the company has used the thing. If it works, this is a footnote. If the typing is bad or the network is the problem I think it might be, Apple will have damaged its actual business to ship a disappointment.
The Mac side is not in bad shape in the meantime, which softens it. The Intel transition is done, the machines are genuinely competitive on performance for the first time in years, and Tiger is a mature system that nobody is desperate to leave. Delaying an operating system release hurts most when the current one is failing people, and it is not.
Developers are the constituency with a real grievance. Anybody who has built plans around the new frameworks now has four extra months of holding pattern, and the ones who took the June date seriously enough to schedule releases around it have had their year rearranged by a decision they had no visibility into. That is the cost of a platform vendor treating its ship dates as provisional, and it accumulates.
What I would actually like to know, and will not, is whether the phone was always going to consume this much or whether something went wrong. A planned reallocation is strategy. An emergency one is a project in trouble, and the two look identical from outside until the product ships.
October is not far. The interesting date is still June.