Cone of Uncertainty - is it a myth?

Is the cone of uncertainty a myth? Barry Boehm presented the cone of
uncertianty for software estimation and it roughly states that
uncertainty decreases over the life cycle of development and estimates
become more accurate. (Is that an accurate summary?)

Do we just accept the cone of uncertainty to be true because it makes
sense or is it real?

I like to bust myths in computer science. One of the recent myths
busted is that programmers can not test their own software.

What about the cone of uncertainty? I think it is a myth as well (read,
trying to provoke conversation).

Gilligan wrote:
> Is the cone of uncertainty a myth? Barry Boehm presented the cone of
> uncertianty for software estimation and it roughly states that
> uncertainty decreases over the life cycle of development and estimates
> become more accurate. (Is that an accurate summary?)
>
> Do we just accept the cone of uncertainty to be true because it makes
> sense or is it real?
>
> I like to bust myths in computer science. One of the recent myths
> busted is that programmers can not test their own software.
>
> What about the cone of uncertainty? I think it is a myth as well
> (read, trying to provoke conversation).

I don't it's a myth. Here is my reasoning:

At the start of the project, when we didn't invest any time yet into
analyzing it, we can't really estimate it - estimations are guesses and
certainly quite uncertain.

At the end of the project, we know can "estimate" time and costs exactly,
because, well, we have the real data from doing it. Uncertainty is zero.

Uncertainty probably decreases more at the start of the project, when we
start to understand what the project is about, and gather experience on how
fast the team can deliver value in the first iterations (especially in an
Agile project).  There still may be surprises later on, with decreasing
probability the nearer we get to the end of the project.

So, yes, I think that it looks quite close to a cone...

Cheers, Ilja

--- Gilligan wrote:
> Is the cone of uncertainty a myth? Barry Boehm [...] roughly
> states that uncertainty decreases over the life cycle of
> development and estimates become more accurate. [...]

In a certain common sense, it's trivially true:  The more you know, the
better your estimates can be.  And as time progresses, and you make
progress towards your goals, you should know more.

However it can be false, if the initial estimates are not rational or
well informed.  On the first day of the project, someone can walk in,
knowing nothing about the project, and say, "This will take six man
months, exactly.  We'll hire three people and deliver it in two
calendar months, exactly."  Time will show that reality doesn't conform
to that estimate.  As time progresses, it will become more and more
obvious that reality won't conform to that estimate.  But in that
sense, the cone of uncertainty is true: Your uncertainty about the
arbitrary estimate will go down over time.

Agreed. I think that the cone of uncertainty is not a myth. After your
comments and comments in other groups it seems clear that as we explore
and reveal things then we diminish the uncertainty and increase our
chances of an accurate estimate.

I do think that the x-axis of the cone of uncertainty graph should not
be time. The passage of time does not mean that expierence is gained.

Example, If I am in a room without windows and a solid door and I guess
what is outside that door but I do nothing for six months and then make
another guess what is outside that door my ability to estimate has not
improved. But If I p under the door, and pry open the door a bit and
then close it and estimate what is out there....

So, I think the x-axis should be some measurement of expierence or
something.

Agreed.

How do we get the cone to come to a point as fast as possible? That is,
what are the things that can help us estimate and how can we take those
things to the extreme and try to get our estimates right as soon as
possible? Any ideas?
I like iterative development and I think that is one way.

--- Gilligan wrote:
> How do we get the cone to come to a point as fast as possible?

The cone can't come to a point until the project is done.  Until then,
it's always possible for unexpected things to happen.

> That is, what are the things that can help us estimate and how
> can we take those things to the extreme and try to get our
> estimates right as soon as possible? Any ideas?

Feedback.

I think the XP approach gives you constant feedback, so you'll know if
your estimates are wrong, as quickly as possible.  Then you can do
something about it.

(I think the XP approach tends to be moderately pessimistic, on long
term estimates.  But this is probably a good thing, both because
unexpected setbacks can occur, and it's generally better to under
promise and over deliver, than the opposite.)

> I like iterative development and I think that is one way.
Yes; it give you very real feedback.

The only certain things in life are death, taxes, software maintenance,
and yet another extreme-programming claim.

If the Cone of Uncertainty is not a myth we are still confronted with
the question of which direction it will point on any given project.

Next one has to worry whether or not the cone simply becomes a
prophalactic for the collection of project uncertainty.

Finally, and this is good advice for newbies, when confronted by
management as to why uncertainty is growing rather than abating always
begin dismissing the Cone of Uncertainty theory as a myth - it's what
keeps it alive.

>>Finally, and this is good advice for newbies, when confronted by
management as to why uncertainty is growing rather than abating always
begin dismissing the Cone of Uncertainty theory as a myth - it's what
keeps it alive.<<

If the time line goes from left to right, the cone is widest at the
left and comes to a point at completion.

As decisions are made and features completed those features have to
move from the realm of uncertainty and therefore the number of
uncertain things decrease until the project is complete.

It is conceiveable that as we start each new feature through
development that we consider all possible solutions everytime and
because we have gained expierence along the way we are now smarter and
able to conceive even more possibilities than we could in our ignorance
and thus the number of uncertain things is increasing and the cone gets
wider and wider and wider.

I think the cone comes to the point because of what I said at first. I
do not agree that our ability to estimate the unknown increases. We do
not become psychic as we go along. The interim estimations for things
we have no expierence in are just as bad as the estimations at the
beginning.

What did you mean that it IS a myth? I want to know more.