h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

All true. When you have a planned programme you can manage it to completion. No problems there (except of course that everyone's so bloody useless at PPM, especially when IT is concerned)

The problem I was angling for was the plan itself. We generate one with a view to delivering our project in a timely fashion - yet, mathematics can prevent us doing this.

I'll be happy to say more, but does anyone know what discipline of mathematics we rely on?

The only thing that I can think of is stochastic modelliing, though that may be entirely wrong.


Geggs

I was looking for something a little more general, but stochastic modelling is a fascinating subject in itself (and certainly worth a Brucie Bonus). The fact that that we rely on randomness in some mathematical processes, particularly in financial applications sounds on the face of it just plain weird.

adding

No, not adding. Let me know if you want me to wash this one up...

Is that a clue?

I think the clever ones can finish it off.

More general? Statistics, then?

I'm clawing at the dark recesses of my mind here.


Geggs

Probability?

Our assumptions of the taks durations derive from probability that they will be completed on time, 95%, 97%, 99% I dunno.

As the number of tasks increase, so the probability that the project will ever finish tends to zero.

Not statistics or probability.

As we're into guessing, I'll give my answer (which is brief and off the top of my head, as all my detailed notes are at home), then invite anyone else to add any related interesting facts for more points:

***

The problem with planning is that it has no proven mathematical basis to it. There is no way of telling that when we produce a plan, it will be the most efficient way of carrying out those activities.

We rely on algorithms to select which tasks should take place in what order and which resources should be assigned to them. These algorithms can't be proven mathematically, although we can show that they work for very small projects (like making a cup of tea), by an exhaustive process of checking every possible combinational sequence of tasks and showing that there isn't any better way of arranging them. When you get to a project with a large number of tasks, this exhaustive method isn't practical, as the number of combinations becomes exponentially large very quickly.

The branch of mathematics is known as network theory, a subset of the wider discipline known as combinatorics. There are many similar problems, for example a salesman who needs to plan as short a route as possible when he visits a number of towns. Algorithms exist to help us, but it cannot be proven that they work except by checking them exhaustively. If there were 20 towns in the itinerary, then you would need to check 20 x 19 x 18 x ... x 2 x 1, ie 2432902008176640000 combinations of routes between them.

Having said all this, there is no evidence that any of the widely used algorithms is flawed, so the construction industry isn't going to collapse overnight as a result of this carpet being pulled from under it's feet. Civil engineers are happy to believe in fairies, so long as they wear a hard hat on site.

***

Hope this makes sense. It was a bit of an unfair question in practical terms, but I think that people should always know whether or not something they use is founded on logic.

Feel free to add further comments - I'll score up next week.

Icy

Hmm.

Construction is surely a bit different to travelling from town to town.

It's so obviously linear.

You need to fabricate your piles before you can drive them, and you can't build your pile cap until you've piled, and you can't build your piers until you're pile cap is cured, and you can't lay your bridge beams until the piers have strength and you can't build your deck until your bridge beams are in place.

No artifically complicated network, no fuss.

Try network theory on MB's Sunday lunch. I'm sure raw turkey and a freshly hot oven are exactly what's not required.

The problem is that we will always try to break things down to make them simple, but all we're doing is rendering it in terms where we can convince ourselves that the subset works exhaustively - maybe it does.

The travelling salesman problem uses a different algorithm, but it's the same principle. I could convince myself that it's obvious that if I have the neighbouring towns of Newcastle, Sunderland and Gateshead on my list of 20, then I should visit those three one after the other, and deal with the towns in the south and west at another time. I'm breaking it down mentally into something which I can handle logically - I'm not thinking about the wider aspect of an algorithm which manages the whole process.

B can't eat his dinner until it's been cooked. Maybe it's possible (hypothetically) to eat it first prepare it second and cook it last, but, patently, the reality is that (in real life) most, nearly all, activities are dependent on completion of other activities.

"The problem with planning is that it has no proven mathematical basis to it....We rely on algorithms....network theory, a subset of the wider discipline known as combinatorics"

I wish you had told me that earlier. It might have saved my job, or at least won me a bigger redundancy package

So, what are the scores Icy?


Geggs

Ah yes, I'd planned to do this. See what happened?

It'll be ready Thursday...

Was that last Thursday, today or in a week's time?


Geggs

This one over

Scores please