Potential explanations for that 'outlier' 3:16 by Carpenter in '93.
After moving to Manitou area, he became very familiar with the course. Doing course segments fast in training convinces the mind that they can be strung together for a fast time during the race.
Knowing every rock, hill and turn on the course is a big advantage. Knowing when and where to push and gain time based on your strengths and confidence, and when to just cruise, is an advantage. Knowing your time splits over segments of the course is an advantage. Knowing intermediate splits is an advantage.
Even though much of the Barr Trail is not super technical, there are technical segments. Running those fast involves skill and risk. Train on the course, develop the skills and confidence to take the risks and you go faster. Even ascending at record pace is a risk (go to hard and you could blow up). Train to get in great shape, decide to take all the risks on one day (confidence growing as you make string together each segment at record pace) and not blow up or fall and you have a record.
Once you have the record, and you know it is out there and will not be threatened soon, why take all the risks again? Just run to win and save yourself for other races.
Pounding downhill really fast on Pikes beats up the quads. If there are other races, why beat up the quads when you will win anyway and there are other races ahead? I do not believe there was any prize money at Pikes Peak then, while other races especially in Europe had the incentive of cash to pursue.
Running really fast downhill on the Peak requires you to be an a-hole and blast past and through all the 800+ other runners behind you still ascending the narrow trail. You can yell and scream and hope they yield, but some are to focused and hypoxic to respond. You can just physically force your way through people. The race leader has a more difficult time as they get to 'remind' all the other runners to beware of two way traffic on an out and back course. Ascend in record time and you avoid relatively more ascending runners on the narrow trail near the top, which is an advantage and helps you descend faster. Decide to be a real a-hole and blast through everyone one year, and be lucky enough that you do not fall or get hurt, and you descend faster.
I am not accusing Carpenter of being an a-hole. He seems like a respectful guy. But downhill runners have the right of way on the Peak and your descent time can depend on how forcefully you claim that right of way.
Only 620 runners finished the marathon in '93. That is 200 fewer(25% less people to dodge on the descent than other years.
Pikes Peak is supposed to be a closed course now but it was not always that way. Past winners and record holders have admitted to training on the course to learn where there were advantageous shortcuts. You would have to train on the course and know every rock, turn and shortcut to do this. You would have to rationalize and justify straying from the designated route in your head to treat Pikes Peak like an open course, like many of the European mtn races allow. Its not like there are course marshals out on the Pikes P course monitoring every turn. Save 20 seconds on a single shortcut on the ascent, and perhaps 10sec on the same shortcut on the descent, and you cut off 30 seconds.
I am not accusing Carpenter of doing this and I do not believe he cut the course.
Carpenter seems super competitive, against himself and others. Having crashed and lost in 92, he came back with a vengeance the next year to set a record. The same thing happened when he did the Leadville 100. He led and faded one year, undoubtedly failing to meet his own lofty expectations, but came back in 2005 to smash the record which still stands
Carpenter trained to be fast on the Peak, and had the skill, intelligence and guts to apply some/most of the above in 1993. Combine incredible fitness and a minute here...a minute there, and you have an outlier record that may never be broken.