The system is evolving much faster than our capacity to understand how the system is working in the first place

From Is CBT for depression losing its efficacy? by T. Johnsen & O. Friborg.

Cognitive Behavioural Therapy (CBT) has many advantages for treating depression. Among them, the fact that it's easy to standardise, it's intuitive and it can deliver results relatively quickly (think weeks, not years as some other therapies demand). For many people with depression, it's also far more acceptable than the prospect of taking mind-altering drugs. But now the bad news: CBT's efficacy seems to be declining.

Anything in the field of psychology and sociology is deeply suspect until replicated multiple times by multiple sources.

But this research does raise the interesting conceptual idea that perhaps one of the challenges related to psychology is the nature of the area of study itself. In other words, human systems are subject to complex self-adjustments arising from multiple different and hard to identify sources. The implication, potentially, is that the rate of change of the underlying human psychological system may be broader and faster than the ability to robustly identify sources of cause and effect within that system.

An early canary in that particular coal mine is the Flynn Effect, the measured trend of an average increase in OECD countries of an increase in population average IQ of about three percentage points per decade. Originally it was accepted that IQ was essentially biologically determined and not susceptible to change over time. Once James Flynn introduced his empirical evidence that average IQ was subject to change, there was long discussion as to whether the trend itself was real. I think there is now fairly wide agreement that it is real. What remains open is the causal mechanism for that increase. And now, even before we understand the mechanisms that led to the increase, it appears that perhaps IQs have topped out among some of the OECD countries.

In other words, the system is evolving much faster than our capacity to understand exactly, or even roughly, how the system is working in the first place.