I think I've seen that take from you before dophin, & you're not alone... That's the mis-assumption people make about 40-times & the NFL. But yes, the idea that "times are at least relative to one another" can be true if you look specifically at the COMBINE (& then only for one given year as a comparitive window). However, beyond that all bets are off.
Consider college pro days like the 2004 UNC one cited in the case of Willie Parker. Remember, at pro days you have literally between 30 & 100 people taking & recording their own times... every scout for any of the 32 teams who shows up, the hosting school, the school of the student athlete running, the agent/support staff (ISP sports) of the athlete running... Do you think it's a coincidence that someone at the pro day for Parker's school reported an astronomically low 4.2 40-time? I'm not saying that Willie Parker is not fast... he may be the fastest player in football without pads-- I doubt it, considering all the BCS conf All-American track stars who play in the NFL, but I have no certainty on this. Rather, I am just saying that this time is almost certainly not relatively comparable to other players measure... even on that day. And of all the heats I've seen for Parker (including campus times when he was an underclassman), none approach 4.2.
There are sometimes as many as 60 or 70 annual pro-day workouts held at colleges all over the US; each with different track surfaces, wind & weather conditions, media/NFL scout coverage etc... But even with all those options, not every student gets to run at his home school-- for instance, while both Virginia Tech & Miami would host their own, other ACC schools like Wake Forest, NC State & regional players like App State, ECU, Coastal Carolina, etc would share with UNC or another big hoster. Clemson/South Carolina (traditional rivals) might share a pro day during a down FB year for one of those schools... in these cases, the home students virtually always have better "reported times" than the comparable visiting athletes, when measured against their combine runs just weeks later.
Sometimes there are even independently contracted COMPANIES timing at pro days, whose sole purpose is to train NCAA FB players to prepare for the draft by improving their pre-draft measure-- these companies understand that a perceived difference of .1 seconds in 40-time can make or break millions in guaranteed money because of the scientific attiude taken by the NFL toward the draft; things like the draft valuation chart. Don't you think they want to falsely report times lower or-- more common-- intentionally take measures using a process that generates faster times?
And there is no SOP between these groups & in some cases even within one of these groups. Even if you exclude the outsiders & just look within the NFL scouts of 32 teams... some of these teams take:
--the fastest time
--the slowest
--the mean
--the median
Some even
throw out the high & low times & then take the mean or median... Sometimes they publicize the results from these methods or from the OPPOSITE of their methods before/after the draft to try & move a player up or down relative to their own or another team's drafting position, to gain leverage negotiating or HURT another team's leverage negotiating with players already picked, to spin public opinion if ESPN reported a GM's selection as being particularly terrible (see Phil Savage's post draft
PR).
It is just a ridiculous circus of misinformation & jockeying among these many forces & almost none of it is believable. You are correct that one given year of times at an Indianapolis Combine is the best source for relative comparable accuracy... anything beyond that, however, is a crapshoot