There are factors involved in Keino's performance that are not being addressed by commenters.
1. Although it is established that there is a trade-off between the oxygen-reduced effects on a runner at altitude on the one hand, and reduced air resistance on the other, there is no evidence that this trade-off resulted in the latter cancelling the effect of the former in any event over 800m. Doubell (Australian) equalled the world record for the 800 at Mexico, closely followed by Kiprugut - Kenyan of course. The earlier rounds had been similarly fast. Clearly, altitude was not a factor in limiting performance at that distance. But in the longer aerobic events we see that every athlete - bar one - ran significantly slower than sea-level times. From the 1500 up, reduced oxygen was not compensated for by less air resistance. This is unarguable. The 5k was run in a time nearly 50secs slower than the wr, and the 10k was won in a time 1.40secs slower than the then wr, and by an altitude-trained athlete.
2. It is being maintained that Keino possessed a freakish capacity in that he was not affected by altitude in the way other athletes were. Freakish, perhaps yes, but it is again undeniable that he was affected by altitude. His relatively poor performances (for him) in the longer events cannot be explained away by saying he was not a specialist at these. Before Mexico he had been a former world record holder at 5k and over 10k his best time was one of the closest in that era to Clarke's world record. He was as affected by the conditions as any of the other top athletes in the longer events, otherwise on the basis of his best performances he should have demolished those felds as he did in the shorter 1500. He was nowhere near his own best time in the 5k and didn't even finish the 10k in a slow-paced race.
3. If Keino was was clearly affected by altitude in the longer races what can be the scientific basis for saying he would not have been affected at the shorter distance, when every other altitude-trained athlete is, and has, been affected, because of the prolonged aerobic nature of the event? If the best middle-distance Kenyans of the last half century (and there have been quite a few) have not run their fastest at this distance at altitude in Kenya how much more unlikely must it be that any could achieve their fastest time at the much higher altitude of Mexico, of over 7000'? It is without any basis in logic and our knowledge of physiology that a distance runner would clearly be affected by altitude in one arerobic event and not another. We do not have an equivalent in any other altitude-trained runner before or since. Keino cannot be a "freak" in one event but the same as his other altitude-trained peers in the others, when these distances were within his racing capacity. He was not a Ryun or Snell trying to run 5-10k.
4. I realise I am discussing these issues with some highly-informed commenters, but I get the impression most of you were not around to see these events in '68. I was. Unless you were there it is hard to grasp the magnitude of Keino's performance. Throughout much of the 60's Elliot's 3.35 was held in awe. No one got close until that incredible 3.33 run by Ryun in '67. Of course there is no automatic conversion table to today's performances, but those times in that era were held in the same regard as we do the records of today. That does not mean that the late '60's runners could run today's times - I am not saying that. Keino's run at Mexico in some ways was better than Ryun's record, given that it was run at altitude. So, viewed in that context, Keino's run in '68 would be like seeing a runner today coming within a second or two of El G's wr - and running it at high altitude. I wouldn't believe that, and, I suspect, neither would many of you.
5. Keino wasn't even the best or fastest man of his era - at either of his favoured events. When he ran his best mile of 3.53 it was exceeded by Ryun's best of 3.51. He consistently lost to Ryun until Mexico in '68. In the 1500 the differential between them was again almost 2 seconds - 3.33 to 3.35. When Ryun was past his peak Liquori became world number one - not Keino. In 1971 Keino ran the fastest time in the world that year of 3.54.4. (Wait a minute - that's the same time Snell ran on grass in '62!) It's good but there is nothing freakish about it. The following year Keino lost in the 1500 final at Munich to Vasala. Again, in his typical region of 3.36 high. If we look at the 5K, his other specialty, we see that although he once held the wr at 13.24 this is nowhere near Clarke's subsequent 13.16, and Keino never won Olympic gold at the distance. His best run cited here, of his 3k wr, was at a distance rarely run at the time - most races were still imperial. The 2-miles was the more commonly favoured race. Others in that event were faster than him. Jazy. Keino was a great runner - chiefly because of his versatility - but there were undoubtedly better and faster in his era. He was no freak. Yet the argument defending his Mexico performance somehow requires maintaining that he was. But only at altitude (and that claim, as we see above, can be dismissed).
6. If Keino had possessed a stand-out aerobic capacity at Mexico it could have possibly - and I say possibly - been achieved through blood-doping. In the late-60's there were allegations the Europeans especially were using it (it had been developed much earlier); claims that became widely publicised in the 70's with Viren. Keino could have used amphetamines if he wished - they were widespread at the time - and so was synthetic testosterone. The means to dope was there. Although the authorities chose to ban stimulants in '68, testing was in its infancy. Rumours abounded of doping at Mexico yet only one athlete produced a positive - for beer.