I don't really think there is much to this. Yes, techniques and fitness and regimens have improved over the last century, but obviously there is a logical upper limit to where that will ever take us, accepting that it explains the long-term increase in standards.
But the fact is... let's say all contestants in track events in the last 16 years- all the TOP contestants- are roughly as good as each other. Every time they run, say, the 100 metres, they do as well as they do- sometimes they do well, sometimes they do badly. There are a million factors that decide how well they do, biut in the end they have got a time, and that dictates how well they have done.
Now, if one person runs that race a million times, the variety in times he gets would be notable, and there would be a top time and a bottom time. An athlete does not run a million races in his lifetime but in this theoretical situation, the best time he gets in those million races is most likely to be a new world record- a one in a million chance. Probably his top 100 times, in fact. So he is capable of doing that.
Now look at the amount of world class athletes running these races in major competitions at least once a month, for every year, over and over and over. Statistically speaking, you'll get more and more people 'lucky' enough to hit the top end of that range of a million theoretical race times, and so more records being broken. And because they are broken in such small amounts- hundreds, or even thousandths, of seconds, the logical point at which that has to end... is probably quite some way off.
Now multiply that out to every single Olympic event and we can expect records to be broken for a damn long time to come!