[SystemSafety] Collected stopgap measures

[Steve Tockey]

Derek,
Of course I wish I could have built several of exactly the same system in parallel. Unfortunately I have only been able to do that once. You realize it is too expensive to do that in the general case.

The next-best alternative is to have by-analogy comparisons. Consider an organization that has built several systems of Type X before, whatever Type X happens to be. So when they are given a new system of Type X they can estimate it based on their past experience.

When we implement that same new Type X system using an engineering approach we, as I said, finish in about half the time, with about half the cost, and with no more than one tenth of the delivered defects when compared with the mainstream estimate.

Keep in mind that this is completely ignoring the fact that the average mainstream software project overruns its schedule estimates by about 42% and its cost estimates by about 35%. We are outperforming the original estimates, not the inflated typical actuals.

Further, for systems of Type X, the organization has experience in maintaining those systems over time. When we compare the maintenance staffing needs on the engineered products to the mainstream products, the engineered versions require between one fourth and one eighth of the maintenance staffing (therefore, maintenance costs).

So no, I don’t have lots of data where the exact same system was developed by multiple teams having various levels of competence.

I have one same-system built in parallel by two separate teams, one mainstream and one engineered. It was clear that the engineering team had a positive influence on the mainstream project. In discussions about requirements with the domain experts the engineering team would ask requirements questions that the mainstream team never even thought of asking. The next time the domain experts met with the mainstream team the domain experts would say, “You didn’t ask this, but here is a very important thing you need to be aware of . . .”

All of the other experience is with the same KIND of system being developed in mainstream vs. engineered approaches where the engineered results were compared to best estimates of mainstream results based on past history.


— steve 


发自我的 iPad

> On Nov 16, 2018, at 6:33 AM, Derek M Jones <derek at knosof.co.uk> wrote:
> 
> Steve,
> 
>> Could that same software have been delivered faster, cheaper, and with
>> higher quality if it had been built using a professional engineering
>> design process?
>> The real-world data I have says yes.
> 
> You have data where the same system has been developed by multiple
> teams, having various levels of competence.
> 
> Great.
> 
> What was the percentage difference in costs and development time?
> 
> -- 
> Derek M. Jones           Software analysis
> tel: +44 (0)1252 520667  blog:shape-of-code.coding-guidelines.com