The absolute best plan (ABP) is a logical construct that is generally impossible to actually achieve, although if cost is no object one can come as close as one desires to the ABP by thoroughly analyzing the work requirements, rehearsing the activities and having large quantities of staff and equipment and materials so that there are no logistic delays.

Since this is generally impossible, the ABP is a theoretical construct against which you can compare what your ActualPlan achieves.

To show that the ABP is a realistic, if rare, entity -- a Illinois based Architecture research group once built a house completely finished outside and inside is appreciably less than a day as a tour de force of planning.

The Empire State Building was built in a single year and Liberty ships were build in as little as twelve days during WWII -- so there is such a thing as ABP. I might add that the "absolute best plan" has to be interpreted in the context of WellFormedWorkPackages because it is only the absolute best plan for that work definition ... that is the plan free of logistic delays and resource limitations.  The "best" part has nothing to do with whether the plan is right! That is the presumption if you have achieved well formed work packages.  --RaySchneider

P.S. A good many points made below are very valid, especially those about "how do you know the plan is any good to begin with ... sorts of things" The underlying presumption in "absolute best plan" is that you know what you want and at least in principle how to get there.  The fact that that is not in general true is a measure of the 1) limitations of imagination, and 2) uncertainties of R&D and risk management is obviously a part of the answer.  Planning is after all a delphic endeavor unless you've done it before, so "of course" the absolute best plan is just the earliest completion based on all resources being available when they are needed ... calling it "best" is just a little emotional overloading.  --RaySchneider
-------
Just to clarify for myself, AbsoluteBestPlan by this definition is the plan that results in the MinimumElapsedTimePlan (as a synonym for AbsoluteBestPlan). Are there other definitions I am unaware of for the other attributes of a project? Examples would be MinimumCostPlan, MaximumQualityPlan, MaximumFunctionalityPlan etc. --PeteMcBreen
----
AbsoluteBestPlan requires AbsolutelyStableRequirements.  Essentially unknown in the software business.  One might consider a more, um, iterative approach to a real project?
--RonJeffries
----
I think you could use ABP as a benchmark in retrospect for software. That is, you get to the end of the project and you say, "Knowing what I knew then, if I had made an ABP, how long should this project have taken compared to how long it took using whatever planning process I really used."

My gut belief is that ExtremePlanning or EpisodePlanning will asymptotically approach an ABP, while more elaborate planning procedures will fall far short, because so much more time is spent planning.
--KentBeck
----
The first definition of AbsoluteBestPlan sounds like "the earliest date by which you can't prove it won't be done by."

Getting stuff done fast doesn't mean you had an AbsoluteBestPlan; it just means you got stuff done fast. If you got stuff done fast, and you successfully predicted when it would get done, that sounds more like an AbsoluteBestPlan. (If you wrote a brand new operating system in six weeks, but your plan called for you to write it in four weeks, what does that say?)

Sadly, the "generally impossible" part also sounds like too many of ''my'' project plans, which don't capture all the deliverables, don't capture all the dependencies, and don't reflect reality more than a couple of days after I put the plan together. (Have I mentioned recently that I hate ProjectManagement?)

Kent's description may sound like something that can't predict when the project will end until it's ended. That may be literally true, but it also sounds like the chapters on Estimation and Scheduling from the book RapidDevelopment, where estimates have error ranges ("15 months plus or minus 3") that get smaller and smaller as the problem is better and better defined.  --PaulChisholm

ExtremePlanning estimates have error ranges, but expressed in functionality, not time. "We will finish December 15, plus or minus some of the stories." That's one of the reasons you do the most important stuff first, so you don't minus much value if that's what it comes to. I swiped this idea from the ScrumProcess.

A lot of the commentary on AbsoluteBestPlan misses the point -- the idea is that any conceptual view of a problem has an AbsoluteBestPlan associated with it in which all the constraints on resources and ideas have been removed.  Obviously if the requirements change, the plan changes as well.  One of the general problems with planning as commonly done, is that it is only done once and then frozen in concrete regardless of the run of events.  That's just crazy!  All plans must be maintained.  The AbsoluteBestPlan is an instance of a constraint imposed by logical conditions.  It is related to the Baby In One Month Syndrome.
-- RaySchneider

I have noticed that the steel framing for an large building seems to go up very quickly and the interior finishing takes forever. Is the AbsoluteBestPlan for steel framing known and followed, unlike the rest of the construction?

No, the pod comes before the peas! The Framework before the essence. Much more detail and construction activity takes place in finishing an interior than does the lifting, positioning and bolting or riveting steel members together. 
----
CategoryPlanning