My previous public criticism of the concept has focused on terminology issues. "Niche construction" is defined by its proponents to refer to environmental modification by organisms. This covers constructive and destructive activities. This is counterintuitive and confusing. A more important concept to me seems to be "environmental modification". Does it matter to the organisms it affects whether the landslide was started by a mountain goat or by a meteorite? Not so much. Yet one landslide is "niche construction", while the other is not.
Part of the idea of niche construction is that it is an alternative evolutionary force to natural selection. Supposedly, natural selection involves environments affecting organisms while niche construction represents organisms affecting their environments. However, there's a problem with this idea. The environment of organisms often consists of other organisms. So, from the perspective of one organism, an event would be niche construction, while from the perspective of another organism it would be natural selection. This severely erodes the rhetoric about niche construction and natural selection being different evolutionary forces. In short, the organism-environment split is subjective. One creature's environment can be another organism. From their perspective, first creature is part of the environment.
Traditionally another area of biology covers interactions between different organisms - symbiosis. There's the concept of a biological interaction, which covers the ways in which creatures can interact. Evolution involving such interactions is known as "coevolution". Rather than having scientific concepts based around the organism-environment split, which is highly subjective, an alternative is to use the existing concepts of symbiosis and coevolution to handle interactions between organisms, and then expand the idea of Darwinian populations so they completely tile the universe. Traditionally, evolution treats a set of organisms and their environment. Coevolution theories show how to deal with parts of the environment that are composed of other creatures. Universal Darwinism extends the idea of a Darwinian population to include practically any set of things. Rocks, atoms, planets, etc can all be modeled as being Darwinian populations. This allows the entire universe to be tiled with Darwinian populations and modeled using coevolution theories. That eliminates the need for modeling the environment separately. The environment becomes just a bunch of other Darwinian populations that organisms can coevolve with.
This strategy of demoting the concept of "environment" eliminates the problems associated with the environment being a subjective concept, that depends on what organism, or set of organisms is being considered. Subjective science isn't necessarily bad, but you have to be careful to make sure that the sums come out the same way for all observers. If A is an organism and B is its environment, models should make the same predictions as if B is an organism and A is its environment. Using separate theoretical categories for A and B increases the complexity of the model and increases the chances of these two modeling perspectives producing different predictions. Coevolution models avoid this problem by treating A and B symmetrically - as individuals in coevolving populations.