I assume you are talking about this graphic for the 13 hour test:
\
I've seen this in a couple of forums recently since Dylan's video. It should be noted that this isn't showing the chains themselves getting faster or slower over time, but is really showing a combination of the lubricant getting faster combined with the lubricant holding ability (or lack thereof) of the chains. When they run this test they run it until the friction hits a peak minimum break point and then begins to increase again. As this was all done with wax based lubricants, you see that the wax gets faster as it's compressed, but then eventually gets slower as the solid lube is ultimately pushed out of the chain. The real takeaway should not be that X gets faster over time while Y gets slower but rather X holds wax lubricant longer.
In general we see that chains without fancy surface coatings hold lube better and longer, this is why SRAM second tier chains like the XO1 and Force test better than XX1 and red level chains, although SRAM has worked very hard to fix that. Though in this test the KMC Gold is a crazy outlier, my experience has been that coated KMC chains typically hold lube worse than most other brands, but most of my work has used the really fancy DLC coated ones.
We've built a rig with Purdue University which is at the moment the most advanced rig in the world to test this. We will be publishing stuff over the next year+ based on the learnings in the lab, and our goal is to ultimately help Purdue become to chain friction what Virginia Tech is to helmet testing, a fully independent lab focused on publishing for the public good.. In the mean time we're using it to not only compare lube and chain but to really try and understand the mechanisms involved here, like why does the friction reduce and then climb in these wax tests, what makes on surface or coating better than another, etc.. What I can tell you from our learnings so far is that every chain and lubricant combination will go through a period of initial improvement over the first few hours and at some time (almost always less than 20 hours) it will hit a break point where things break down and friction begins to climb again.
This break point and climb is an indicator that you have metal on metal beginning to occur and that wear has begun, so you really want to avoid this regime of the curve.
If you catch the chain before this breakpoint and 'reset' it as Adam likes to say, you can repeat the same cycle over and over again hundreds of times with essentially no wear almost regardless of what chain you are talking about.
If I was after zero wear, I'd look at this chart and pick my chain and make a plan to 'reset' it before it hits the breakpoint, you'll easily get 24-30k km from a chain treated like this.
If I was after the fastest chain, I'd look at this chart and buy the chain with the lowest peak minimum friction and then time it such that my event occurred in the Z amount of time before hitting the breakpoint
http://www.SILCA.cc Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc