Yeah the problem with excel based processes isn't really reliability (not the least because they are built and maintained by users who understand the domain rather than programmers writing code on specs), but lack of automation.
It's also very inefficient. Sort of a push approach where if you change an input you need to recompute every single combination of outputs instead of just the one output you want, as opposed to code which has a pull approach, where you call your output and it recalculates the steps it needs for just that output.
And as soon as you get into something you would express as a loop, or loop of loops, excel starts to be really painful to use if hardly usable.
Comments
If you find a way to audit spreadsheets to find fraud and error you'd make a fortune.
Speadsheets are scary.
http://www.eusprig.org/2006/spreadsheets-in-clinical-medicin...
This spreadsheet "shaming" is getting tiresome. The sheer amount of spreadsheets means that more than a few mistakes will come to light.
Let's not forget the fMRI software with bugs. Was it done in Excel?
Sure, Excel has some problems. No, they are not even close to what is implied in HN threads like this one.
Yeah the problem with excel based processes isn't really reliability (not the least because they are built and maintained by users who understand the domain rather than programmers writing code on specs), but lack of automation.
It's also very inefficient. Sort of a push approach where if you change an input you need to recompute every single combination of outputs instead of just the one output you want, as opposed to code which has a pull approach, where you call your output and it recalculates the steps it needs for just that output.
And as soon as you get into something you would express as a loop, or loop of loops, excel starts to be really painful to use if hardly usable.