A spreadsheet too far?
All readers are aware of the convenience of spreadsheets for storing data. Moreover, tabular representation can reveal by visual inspection, this supplemented by some basic calculations and easy graph/chart construction, key features relevant to decision-taking. Appeal rests firmly with how readily spreadsheet technology enables people to do many tasks without recourse to others with specialist skills.
Downside arises from the restricted view of data analysis imposed by spreadsheets. it may be encouraging many to seek increased analytic power by beefing up spreadsheets rather than looking for simpler to use alternatives. Additionally, other than for simple analysis spreadsheet formula use, built in or now bespoke, is a messy approach to programming. I hazard that complicated cross-referencing formulae embedded in cells is not only difficult to set up but also error prone.
Thus, instead of continuing with increasing spreadsheet complexity their distributors, commercial and open source, might offer better service by complementing spreadsheets with easy to deploy software that abstracts necessary rows and columns for analysis. Perhaps sometimes the results can be returned to the originating spreadsheet but oftentimes they may better be displayed by other means; statistical analysis beyond basic descriptive statistics is an example of the latter.
The great advantage of abstracting data for analysis rests with the code being easily re-used and adapted. It naturally encourages good coding practices.
Obviously the approach mentioned above is already in widespread use. When databases of greater complexity than spreadsheets are handled it is necessary to abstract desired collections of information as raw input to other programs; note that in the minds of people drawing from complicated database structures there is no inclination to believe embedding analysis within the database, even if feasible, offers advantage. Also, some statistical packages in common use, e.g. SPSS, store data in rows or columns, give users opportunity to derive further rows and columns, but don't make analysis procedures part of the notional cells in the data store.
It seems time to recognise that spreadsheets have gone further than desirable as combined data stores and analytic engines.