>
> "Some people familiar with R describe it as a supercharged version of
> Microsoft's Excel spreadsheet software..."
>
It is easy to ridicule this line from the NYT article. But this is not only
a very sensible comment by a smart reporter, but also one that is good for
R:
It is good for R because it explains the new (R) in terms of the familiar
(Excel). Of course R can do far more than Excel ever could, but most
readers will not be familiar with boxplots, let alone studentized bootstrap
confidence intervals, yet R is useful even for elementary analyses.
It is good for R because it will bring us new users. I have often looked
over the shoulders of Excel users struggling to do analyses or construct
graphics that are just slightly beyond what Excel makes easy. Perhaps the
dataset is too large, or the analysis doesn't fit into the spreadsheet
model, or the analysis isn't built-in (and so requires either many manual
steps, or Visual Basic programming, or an expensive add-on package), or it
requires data sources that Excel doesn't handle well, or it has gotten so
complicated that it is unmaintainable in spreadsheet form. R scales better
in every way: in size of problem, in complexity of analysis, in data
sources.
It is good for R because it makes it sound unthreatening and easy, both for
the person who might consider using R rather than Excel, and for his/her
management. Of course, R is not trivial to learn, but you don't have to
master everything about it to get useful results (just like Excel, I might
add).
It is good for R because it reminds us that there are other useful computing
paradigms that we can learn from. The spreadsheet model, including instant
update, is compelling for a wide range of problems. I have not used any of
the R/Excel interface packages, but presumably they combine the advantages
of the approaches. Perhaps there is room for not just integrating R with
Excel, but for incorporating the core ideas of Excel into R in some
intelligent way.
It is good for R because it shows areas where R can be improved. Excel
makes it very easy to present tabular data and format it. It makes it very
easy to work with summary/contingency tables (pivot tables) interactively
and only a little more difficult to do drill-down. In all cases, its
functionality is limited, but what it can do, it does well.
It is good for R because it reminds us that there are many people using
other tools who could benefit from outreach from the R community, both
through tools (smoother interoperability) and through education.
All in all, characterizing R as a supercharged version of Excel makes a lot
of sense.
-s
[[alternative HTML version deleted]]
