[Chandler-dev] [Sum] The Great Architecture Discussion of 2007

[Philippe Bossut]

Hi,

Phillip J. Eby wrote:
> On any *single* criterion (performance, scalability, testability, 
> complexity reduction, etc.), you can easily come up with one or more 
> proposals that will achieve some improvement on that criterion at a 
> lower cost compared to the *whole* of what I've proposed.
>
> But no counter-proposal made so far can match mine on *all* of the 
> criteria, nor can any *combination* of those counter-proposals match 
> mine in overall cost-benefit ratios, after you add up their individual 
> costs.

Hmmm, may be I missed something in the thread but I haven't seen numbers 
related to "cost" flying around so it's hard to add them and even harder 
to compare them. So it's a rather hollow argument.

It's not true anyway that final cost is the only criteria. For instance, 
if we had the whole set of numbers, we *might* choose to spend more 
overall (over 6 years say) and go with a path that provides more 
predictability or a lower burn rate. We may also decide that the overall 
cost is way too high under any hypothesis and decide to drop some of the 
criterions.

All we can say right now is based on "gut feeling" rather than data and 
research. Not shameful (that's the nature of business) but worth 
pointing to. Trying to clear some of the uncertainty with a limited 
pilot project (which we could use as "research" to qualify the "gut 
feeling") seems to be the only logical next step here.

Cheers,
- Philippe