Many products fail before they truly become obsolete though, which is the result of cheap component choices in many cases, which as described in the above article is a very real dynamic to modern day manufacturing based on lifetime projections. Plus many items simply don't go obsolete based on visual aesthetic (TV's) or performance/crunching (computing), as you partially mention above, objects in which performance gains are small or based on pure subjective opinion, such as receivers/amplifiers, speakers/subs.. microwaves, refrigerators, ovens, etc, but still they fail due to deliberate design in many cases. Which is the case for deliberate planned obsolescence.
FWIW Intel's Core i5 architecture debut in late 2009, 3 years ago. In that time their have been 2 small architectural changes to the process, but they still have the same basic and efficient structure. Benchmarks put the new Ivy Bridge revision (2012) as maybe 35%+- more efficient than the Lynnfield Core i's 3 years past. GPUs which are becoming more general purpose, and thus more important in everyday computing, have also seen relatively small improvements over the last three years. Which is simply to say some years are better than others for technological jumps. And that we don't always see a doubling in (consumer product) performance every 18/24 months as a Mr. Moore implies in his oft quoted law.