Fantasy IT strategy?
More like 28 weeks maximum.
I like Peter2's idea. It's clear *some* projects must becoming out on time and budget, if only because they have started so many of them.
<rant>
But what will they show? The usual suspects of badly thought out ideas, backed up by inadequate analysis (with little or now input from actual front line staff) "managed" in a desultory way by a committe headed (usually) by the junior-assistant-under-under-under-secretary. The supposed chair has got more important things to do than manage an umpteen million pound project when he could be playing golf.
Naturally the requirments will move in unpredictable ways on a regular basis and none of the management team will have *any* feel for the capabilities of what their current systems can do (which might be quite extensive) or the limits of what future systems could do (store details of every single email, text message and land and mobile phone call then pattern match them looking patters near that of an actual terrorist's comms pattern, for example) .
The work will of course be knocked out to one of the usual suspects whose fine body of work has graced these pages on a regular basis since I started reading El Reg. Probably the one who has f%^&ked up the least recently, or one who bunged the Minister the most.
They in turn will knock it out to their offshore subsidary who will *atttempt* to make sense of a project plan put together with about as much throught as Dr Tim Mitchell seems to have applied to the architecture of the Climate Research Unit's suite of "analysis" software.
The software will finally be delivered witht he usual cost and schedule overruns to staff with effectively zero training (because of course its so simple to use it doesn't need any training)
This will generate an avalache of Change Requests which the Government (or its taxpayers) will be charged through the nose for.
Unless the new bunch of cretins (the musical chairs of senior Civl Service moves will ensure that nobody from the original batch will still be responsible for this) decide to either end development and start a new system or (unthinkable!) call it a pile of cheesy c&*k and abandon it.
</rant>
I note large, complex, safety critical system with fairly hazy initial specs have been delivered on time and on budget (but it's quite a big budget). The space shuttle on board systems being one such.
The basic principle seemed to be build a framework then add (extensively) tested modules to it as your understanding of what is needed grows. NB. Honest developers have admitted that getting that designing that initial (working ) minimal framework is *hard*.