Excel graphs, like PowerPoint, are optimised for the “sales pitch”. Optimising it for another task is a hassle, and I’d suggest that using a product with a different focus can save a fair bit of that hassle.

Even if this refers to the “personal computer” and not general-purpose computers, I don’t know that it’s correct. See the Computer History Museum’s Timeline at http://www.computerhistory.org/timeline/?year=1963

Decide if the level of eye-candy your audience needs goes beyond what you are willing to add.

Nobody *needs* any type of candy. Rotting teeth and rotting graphics are good discouragement for the candy cry-babies.

Your point 6 is particularly valid when there are a lot of people pushing Tufte’s work, without really considering the reality of creating visualitions in a business environment.
I think the challenge for businesses is to produce adaptable templates for their common tasks, realistically (and often unfortunately) using Excel and PowerPoint.

For templates, this means simple – I’d take your point on there being no hard and fast rule against clipart or 3D text, but would argue that a template must be simple to work under the most circumstances possible. Most of the time, 3D means more clutter, so starting with it is bad.

Thankfully, one thing Microsoft got right in Office 2007 – alongside all the things they got wrong – was shared chart templates. Moving data from Excel to PowerPoint, which we all have to do, gets a little bit quicker and you can spend a little more time on elegant visualisations.

I would accept some ‘fluff’ to evoke an emotional response in graphs in newspapers and magazines and such documents. As you say in such cases, the viewer may just turn the page if there are no attractions – but in business – I am not convinced.