JD_Toims - 12:50 am on Dec 16, 2013 (gmt 0)
Thanks johnhh -- I agree about not wanting to get into an argument about it, so I'll just leave it with:
If someone's storing "critical data", such as in a bank type setting, then by all means, forget about speed gains and normalize absolutely everything.
For most websites though that's not necessary, and since even 0.2 to 0.4 seconds can make a difference in visitor use/satisfaction* I "err on the side of flat-out speed", which, again from my testing, means throw out the normalization and go with a "big table" design almost every time.
* For a source on the importance of speed, see Google's test results from a few years ago when they added delays in showing the results to their visitors to see if page load speed is a factor wrt visitor behavior and satisfaction -- I don't have time to track it down right now, but IIRC 0.2 seconds made a noticeable difference in use and 0.4 second made a fairly large difference in use.