File size comparisons, InDesign CS3, CS4 and CS5?

Question

Hi, all.
It seems there was a trend for several major releases where each time Adobe released a new version of InDesign and InCopy, average file sizes grew by 20% or so from the old release to the new release, at least back in the older CS days. Has this trend continued, such that file sizes in CS4 are substantially larger than file sizes from CS3, and CS5 files are substantially larger than those from CS4?
Adobe, of course, wants to keep writing functionality that will keep the user community coming back to buy in to upgrades. The added functionality sometimes comes at a cost beyond the price tag. If file sizes are larger in a newer version, then page saves over a network or to a database are likely to be slower, and user productivity takes a hit while users are waiting for files to be saved.
Has anyone done any testing to build the "same" page in multiple versions of InDesign to understand what the effect is on the file sizes? I'm specifically interested in the file sizes between InDesign CS3, CS4 and CS5. To be meaningful, the test page would have to be at least moderately complex, with a couple of photos, multiple text elements and so forth. By "same," the page wouldn't take advantage of new functionality in newer versions but would be saved as a native page of the current version, though the file sizes may be bloated by the new functionality like it or not.
If you've done any testing along these lines, I'd like to hear more about it.
Thanks.
   Mark

Answer 1

The overall structure of ID's files have been exactly the same for -- as far as I can see back -- CS. No change at all, in there.
There have been significant additions to the 'global' spaces; stuff like InCopy user data, table styles, object styles, and cross references. Each of these add a major chunk of data to each file, whether you use it or not, plus a few bytes per object (again, whether you use them or not -- ID also needs to know where you did not use them, that's why). I think these might be the main source of 'global' file size increase (a single object style in one of my files, for example, eats up a healthy chunk of 11,482 bytes).
For the rest, all new stuff like 'span columns' is a handful of bytes per paragraph style. Tracking changes may very well double the size of text runs -- but 1 character takes up 1 byte of storage (plus perhaps some overhead of indicating its 'tracking' status). Any single measurement unit always uses 8 bytes at least (for example, the left inset for a column span -- even if it's not used, or set to 0pt).
I think we're talking about a couple of K's here (oh -- perhaps a max of a hundred or so), in a file format that has been designed around the concept of "disk pages", each 4K big, meaning that sometimes adding one single character to a text box increases the file size by 4 K.
Your idea of comparing the size of a file created in CS3 against the same saved as CS4 and as CS5 is certainly feasible -- I might try it some time, just to confirm it's purely the extra 'new objects' data that accounts for the size increase and to confirm my guesstimates of the number of Ks involved.