While they aren't separated, sometimes you have to make a choice of simplified models with certain boundaries because of limited computational power. See also the sequence about reductionism for more on that.
People also did this historically because some categories intuitively seemed more concrete than others. We're moving away from that because those categories have been explored thoroughly enough that we can see the links between them, and which new hybrid categories this points to.
But, yes, you're right, the frontier is moving, and cool stuff awaits beyond.
Again, I'm not saying that conceptually we'll succeed in having no boundaries. I think if I had rephrased it as zooming in on boundaries, it would've been clearer.
The examples that I gave have the property that people didn't even know that these boundaries existed earlier, but in retrospect the collision seems obvious. So that's the kind of examples I'd be more interested in learning. I think the great limitation of not having the language to meaningfully talk about it is the biggest problem in recognizing these implicit assumptions. For quantum computation to develop, you needed (a) Quantum mechanics (1920s) (b) Computability theory (1930s) (c) Information theory (1940s) and (d) Computational complexity theory (1960s). This lead to people thinking about quantum computing the 1980s. So, if the two fields aren't expanding to collide, it's hard to notice.
Would you say the boundaries are getting lower? With increasing specialization, it seems it gets harder to connect an ever greater number of disparate fields (with a couple of choice exceptions, as you note). Of course, in nature there is no boundary (eg., between chemistry and physics), but there are limits to what fits in a human brain.
I'm not saying anything about specialization. When a collision happens, new people start specializing in the collision region. For example, now there are quantum computer scientists, behavioral economists and cognitive scientists working in AI.
I'm not even saying that all fields have to collide. For example, a historian has nothing to do with quantum physicists because there is no boundary to dissolve, implicit or explicit. But the boundary between history and sociology is a different case. I'm not saying: "Yay! Everything is connected!".
What I am saying is that there are (possibly important) connections that we're missing because of our implicit views of reality. I was talking more about realizing that what you think are clean boundaries have to have deeper and more complex structure because of the nature of reality.
The more we ignore the cognitive barriers we have placed upon the world, the greater our ability to problem solve. My own lab group (chemists and materials scientists) recently collaborated with microbiologists, civil engineers, and industrial engineers to produce a feed spacer which blocks biofouling in RO membranes using technology originally developed for use in medical devices using a catalytic process discovered by a nutritionist. It reduces biofouling by 3-5 logs, which is unprecedented, using a process that engineers never could have come up with had they remained focused on their own narrow field.
There is a reason that most universities are now encouraging collaboration between different departments.
It is useful to have specialized knowledge, of course, but we mustn't limit ourselves to our respective fields. Honestly, I think I would be against the creation of new fields of study at this point. Rather, people should be encouraged to work together with people of diverse backgrounds so we can apply what we have learned in one field to others.
As to the question, to pull a Rumsfeld, we have known unknowns and unknown unknowns. I could only guess at where the trend toward broad-based collaboration will lead us.
Molecular biology and physics are in the process of colliding.
Though of course, the changes that I'm blind to, I can't share with you.
A potential boundary is that between the seemingly objective and subjective. At the moment, science seems to be set on measuring things outside of the self. There is relatively little exploration of what one subjectively perceives. This facet of existence is almost completely ignored, even in psychology. I think in time this boundary will have to dissolve. There really is no good way of separating the outside world from the inside world.
Another one could be the artificial boundary between nature and nurture. It seems like the expression of all genes is mediated to some extend by the environment, therefore it is senseless to talk about any boundary between the two. I recently heard a metaphor that the debate between nature and nurture is akin to debating whether length of width is a better determinant of the area of a rectangle.
Computer science and information theory were separate from physics. Not anymore. People realized that information had to be physical and this had profound consequences, especially in the form of quantum information/computation.
Psychology and economics were separate. Not anymore. People realized that humans were the core of economic systems and their behaviors fundamentally shape the nature of economies, even at the largest scales. Note the rise of behavioral economics.
Neuroscience and computer science were separate. Not anymore. People realized that thinking about the brain as a computer is probably the best possible abstraction to understand it.
Reality exists. There are no intrinsic boundaries in reality. All fields of study are created by humans. But these divisions seem so natural that nobody realizes that the boundaries have to dissolve. The fields have to collide. And when we realize that--or finally have the language and ideas to meaningfully talk about it--we find out all of kinds of crazy, cool stuff.
So: what collisions are we currently blind to?