The reason the edge is where the holes must go is that this is the area of highest pressure and lowest velocity. This is where resonant absorbers work best. The Riser absorber approach is a resonant absorber, at least that is my take. It must be, even a leaky box with a small opening resonates. If it was not a resonant absorber but rather a velocity absorber, then the holes would actually be better placed farther from the walls where velocity is highest and pressure is lowest.
The reason small holes are bad is that the tuning frequency becomes very low to the point of being ineffective. Technically if you had enough small holes it actually would be fine. There are ways to calculate it, but it would be a lot of small holes all over the riser. Using relatively long narrow vents is the best option in my opinion, and you really do need a lot of them. As in along every perimeter. When I say long and narrow, I don't necessarily mean 2 x 60 either. Wider is still better, its just that if the riser is 120" wide, then you want as much of that, say 110" of the length, to be open as possible. The wider the vent the larger the port area the better because it raises the tuning into a more appropriate area.
By my calculations on a typical single riser, they often end up being around 1000 liters and if you add in 1000 square cm of port area to the perimeter, that creates a tuning in the 30hz range or so I believe (I'm recalling this from memory, it may have been lower, but certainly not higher). The Q was also very low with such a design meaning the bandwidth was quite wide relative to typical pipe style Helmholtz resonators. If you reduce the opening, then two things happen. First, the tuning drops to an unreasonably low level. It would be reducing bass output in a region where you don't need to add such damping (below 20hz). In addition, it would raise the Q creating a smaller bandwidth and increasing the potential for this riser to actually create a suckout in the bass rather than generally helping to absorb. The insulation will mitigate this however, so I don't want to create the idea that this is a huge concern, its not.
Another thing to consider is that these openings are ports and as such their depth matters. If the top panel is 3 layers of plywood, mdf, subfloor, whatever and you add in a vent that has a depth of say 2" or so, that lowers the tuning. The port then is effectively a 2" deep port. Let's say its 4" by 30" each. The depth is 2". Let's say that the total number of ports is 6. The total mouth area is 720 square inches, or about 4600 square cm. Then with a 1500 liter riser, we end up with a tuning frequency in this scenario of about 37hz and Q of about .3. This makes the bandwidth huge, more than 120hz of bandwidth, but the overall effectiveness small due to the small rise at tuning. Think of it as if the absorption coefficient was something like .4 from 26hz to 100hz, and then falling off after that on either side. Compare that to a high Q design where it might have a coefficient of 1 at 37hz, but where its only .3 at 50hz and 25hz. If we use no diffuser on the vent and just one layer of plywood on top, then we have two problems. One is that the tuning is now over 100hz, too high, and the bandwidth is now a Q of .1 or so, too low. it basically won't do much. Because its a high pressure zone, it won't offer a lot of absorption via velocity either because velocity absorbers are least efficient in this location and the surface area is too small. The top is also now a flexible resonant structure and so the entire top panel between the spans will be able to resonate like a panel trap, but with somewhat unknown tuning and Q. In general the port tuning and panel resonance are not in phase and so what happens is unpredictable and unsmooth absorption. This is essentially what happened in the BBC tests of Helmholtz absorbers, which they abandoned testing. Again, this is my take.
