Choosing a tool

[Uffe]

Hi all,


This is a tangent from Ian's thread concerning EA vs Aris for process modelling.

I realize that 'it depends' is the most logical answer, but, like most clients, they haven't decided exactly how they are going to do their process modelling.
And I think this is a reasonable approach.
But before you scream 'that's no way to choose a modelling tool', given that they have realized they WILL need a tool, then surely the smart thing is to tailor their approach to (1) their own requirements, but also (2) the capabilities of a tool.

It seems like a solution -> Requirement approach, but don't we need some measure of this? No point crafting a wonderful modelling approach, then finding there is no tool to which can do it without huge modification. And when maybe a small change to the modelling approach would make it fit disproportionately better with one or other tool.

I couldn't agree more with this, and in general universally if you refuse to allow the chosen platform to drive requirements you will end up with a complete mess that combines weak least-common-denominator fulfillment of functional requirements with the massive bloat of "platform independence."

It doesn't matter if you're designing a system, devising a method or picking a tool, if you put those blinkers on (horizontal blinkers, I guess, preventing you from looking down) you'll end up spending an enormous effort on shoehorning which, ultimately, is non-productive work.

Understanding the platform (tool, middleware, operating system, programming language...) and its capabilities, and making sure your solution fits on top of it, is Right not Wrong.


/Uffe

[addisonsophia]

Client considerations have to be paramount. Blinkered eyes never help even if that is usually what we have. Work is always a good thing, non-productive work isn’t.” Platform independence” ends up being an abused term. There is no real sense of independence.

[Glassboy]

I couldn't agree more with this, and in general universally if you refuse to allow the chosen platform to drive requirements you will end up with a complete mess that combines weak least-common-denominator fulfillment of functional requirements with the massive bloat of "platform independence."

I've seen quite a few modelling projects fail.  The thing common to them all was taking the traditional approach of gathering requirements, evaluating the tools in the market, and then choosing the tool that best fulfilled the requirements.

What ever project failed to realise is that near universal access to the tool is required for modelling to be successful.  None of the projects had the lowest price per seat as a requirement.

[Paolo F Cantoni]

I couldn't agree more with this, and in general universally if you refuse to allow the chosen platform to drive requirements you will end up with a complete mess that combines weak least-common-denominator fulfillment of functional requirements with the massive bloat of "platform independence."

I've seen quite a few modelling projects fail.  The thing common to them all was taking the traditional approach of gathering requirements, evaluating the tools in the market, and then choosing the tool that best fulfilled the requirements.

Whatever project failed to realise is that near-universal access to the tool is required for modelling to be successful.  None of the projects had the lowest price per seat as a requirement.

Your point is well taken, but with regard to universal access...  I think many modelling projects fail because most people actively (or at least passively) resist modelling.

We (at my workplace) experience this.  We, therefore, concentrate on reducing any friction encountered by our users in using the tool.

Paolo

[Glassboy]

Your point is well taken, but with regard to universal access...  I think many modelling projects fail because most people actively (or at least passively) resist modelling.

We (at my workplace) experience this.  We, therefore, concentrate on reducing any friction encountered by our users in using the tool.

Because - to an extent - modellers are born not made.  Which is why universal access is a prime consideration.  It's not going to the people you think who drive the model forward.  Most modelling tools get wasted on a bunch of opinionated self important architects. :-)