As a quick PS, being able to subtly shift MIDI notes around in time is another reason for not starting at the beginning of bar 1 on the time-line - a few blank bars allows room not only for the MIDI stuff to sort itself out and sync, but also for shifting notes a few clicks (or more) forward in time.

Original Pranksta
I just recorded the stream of the Pat Benetar song mentioned in the article, and used Melodyne Essentials and SONAR Platinum's Region FX tempo mapping to take a quick look:
 

 
I don't think Melodyne does an adequate job of accurately mapping tempos, and the results should not be taken at face value.

 
You can see in the article what my results were:
 

 
Remember that it's essential you use the correct algorithm in Melodyne when doing tempo extraction. I forget whether I used Percussion, Melodic, or Universal...also my screen shots were generated using Melodyne 4. That might make a difference.
 
Tempo extraction isn't perfect, but in my experience it's 98% of the way there. It's easy to confirm whether the tempo extraction was successful and accurate - sync a drum machine to tempo, and see if it follows the track or not. In all the examples shown in the article, the drum machine lined up with the song.

∞

Original Pranksta
I used Universal in my example.

Try the percussion algorithm and see what happens...
 
As to more painless tempo map editing, once a program can shift time without artifacts, we'll be good to go. We're not quite there yet.

∞

∞

Hmmm...I can't explain the differences between your results and mine. As mentioned, I confirmed by driving a drum machine with the tempo map and it matched up with the songs. Maybe it has something to do with my using Melodyne 4? I can't try it with Essential because 4 replaces Essential.
 
As you saw in the article, I tried it with a bunch of different songs so I'm at a loss as to why it worked consistently and well, whereas it didn't for you (unless there's a Melodyne version difference). Bear in mind, not everything I tried gave perfect results - but those were the exceptions, not the rule.
 
Then again, the point of the article wasn't about tempo maps, but to demonstrate tempo variations in classic songs. For that purpose, Melodyne worked like a champ. I certainly couldn't think of any other way to show tempo variations graphically with that degree of detail, so I was very pleased it worked.

∞

These discussions make me wonder sometimes if music hasn't taken a step backwards due to computers. We went from focus on musical content to focus on visual analysis in only a couple of decades (among other issues). It is amazing that music has been made for centuries without computers, yet we seem dependent on them in today's world.

∞