I have suspected for some time that wuzak's "NIH" comment was the central issue on U.S. 20mm development for aircraft. The U.S. Army - and thus the USAAC/AAF - had a near-unholy love of the 37mm gun. Virtually every attack/ground support aircraft under development immediately prior to and during WWII had the 37mm cannon as part of of its potential armament. In many cases it was one of a number of different armament fitments. I have a copy of an original engineering drawing for mounting the 37mm in the Wing of the Mustang, probably for the A-36 or early Mustangs (sorry...can't immediately recall for which).
Nearly all American heavy bombers had 37mm turrets as part of their early design studies, and the AAF did studies for bomber mountings of cannon up to 105mm. I believe it as Oldsmobile that actually had a working prototype for a 105mm short-barreled cannon for mounting on heavy attack aircraft, and the A--26 had one mounted in a gondola in their initial proposal documents.
I suspect that this love of the 37mm gun may have had something to do with John Browning's involvement with the design of the 37mm cannon, but that is speculation on my part at this point. Browning's development of the .30 and .50 cal machine guns certainly gave him very high standing in the eyes of the command structure.
This can't be the only factor in play as the US Army was quite interested in both the 20 and 23 mm Madsen cannons, but apparently mostly rejected them on the basis of cost/complexity of machining the curved components it used. (though the modest rates of fire and lower velocity of the 23 mm may have also been concerns. but the 23 mm being belt-fed was certainly a plus)
Additionally you had the adoption of the Hispano HS.404 based on French specs, but that had continuous reliability and some quality control problems. (the NIH issues may have contributed to that, ignoring British advice on improving it, though apparently tolerances were also too loose due to being relegated to the Artillery category along with everything else over .60 cal ... this may have been a factor for the interest in .60 cal as well, as it would be a bureaucratic workaround for that problem) Presumably the 37 mm M4/T9 was conservative and foolproof enough to work well with those tolerances and was based on a weapon already in use as ground Artillery. (then again, the Oerlikon S did fine in Navy service, so that seemed to avoid similar issues: also another NIH example of German/Swiss engineering/design/license)
Further, while the long-recoil browning action was used in the 37 mm weapons (both the M4 and the high veloicty M9, both based on older ground based autocannons), the US Army, or certain elements of the Ordnance department seemed opposed to scaling up the short-recoil Browning Machine Gun action beyond .50 cal for unknown reasons, so serious developments on that don't seem to have taken place. (the Japanese, of course, did quite a lot of work and scaled it to 20, 30, and 37 mm successfully, though using lighter and/or lower velocity projectiles than the US Army would've likely preferred: the 30 mm weapon may have been of similar size, weight, and rate of fire to the sort of 23 mm high velocity gun the Army wanted, see below)
The Army was also very interested in a high velocity 23 mm cannon and did some development work on that pre-war and early war, but didn't find anything satisfactory:
CAL90 A long recoil weapon similar to the M4 cannon was developed which was considered too heavy and slow firing (and also wasn't a simple rechambering of the existing gun, so not even appealing for rapid manufacturing start-up, ie as a necked-down 37mm rimmed cartridge may have allowed, also making for a legitimately decent anti-armor weapon and maybe assisted in introducing belt-feed with longer, disintegrating link belts rather than the loop type or very short 15 round ones: I suspect the belt-feed system was too weak to pull through heavier, longer belts, but may be wrong there)
I commented on this here a few months ago:
Effect of the US produced HS.404R4M 20mm cannon being perfected pre-war? (see some corrections in the follow-up post below that one)
I also noted the potential of adapting the Oerlikon S (or FFS) to 23 mm by necking the case out like the Madsen did, and the advantages of that specific to a blowback (API blowback in this case) system, with the case being basically straight-walled with little or no bottle neck at that point, more like most other API blowback cases. (taper and bottleneck increases bolt-thrust and thus bolt mass required in a blowback system, which in turn also requires increased spring tension to absorb the inertia of the bolt, and that bolt thrust increase is proportional to the ratio between bullet-diameter area and case-base area ... as such, with case base staying constant, you could maintain a similar muzzle velocity with a constant sectional density of bullet, provided you had sufficiently energetic powder in use: in reality, powder mass adds to the inertia, too, so actual bullet energy is going to reduce slightly as bullet weight increases ... which is mostly fine given the finite case capacity anyway) In any case, a moderate-high velocity 23 mm Oerlikon S derivative seems plausible at least, particularly with a fairly light 23 mm projectile, more like the Madsen used. (174 grams in that case with a MV in the range of 770-800 m/s)
API blowback guns have the disadvantage of needing lubricated/wax-coated ammunition, but that was a known evil at least, and much less problematic than the manual greasing/lubrication required for remotely reliable Hispano 20mm use with the USN (though the heavy greasing may have mostly just accounted for the overly loose chamber tolerances and avoided light-strike problems). Plus, a durable-but-thin wax coating tends to preserve and waterproof ammunition, making it more suitable for Naval and tropical use. (the sort of wax experimented with the .276 Pedersen in the late 20s/early 30s seems quite relevant)
Large guns.
Heavy ammo.
Very short barrel life.
US Pilots were advised to fire only 75 rounds through a cold barrel before stopping to let it cool and only about 25 rounds per minute after that.
Obviously not practical in real combat but a sure indicator that barrel life was not good. High velocity guns funnel an awful lot of high temperature gas through small diameter barrels.
The Germans traded the hitting advantage of the 15mm cartridge for the greater destructive ability of the heavier projectiles. They changed course.
In the German case, they also had war-time pressures limiting development, otherwise the MG 151/20 may not have been a simple rechambering with the barrel change being the primary difference. They may have preferred a somewhat higher velocity weapon had they been afforded the time to start-up new production with a lengthened receiver and somewhat longer cartridges.
OTOH, there's the confusing dichotomy in 30 mm development with apparent continued interest in the very heavy, moderate rate of fire MK 103 (and earlier MK 101) and the low velocity MK 108 without any sort of middle ground. (post-war 30 mm developments with the ADEN and DEFA ended up with just such middle ground using cartridges roughly similar in size to the MK 108's, but with longer cases matched to shorter, lighter projectiles at much higher velocity)
The Germans had also started the war with the modest velocity MG FF cannons and passed over the intermediate-velocity FFL or high velocity FFS. (several countries including France and Britain tried the FFS in some form, but only Japan seems to have taken a serious interest in the FFL, with the Japanese Navy developing it and switching over from initial use of the lighter, lower velocity FFF) Side-note, but the FFL used basically straight-walled cases already, so necking up would not have been an option. (though necking down to 15 mm could be done, and using similar 64 g projectiles to the German 15 mm, velocity in the 840 m/s range may have been possible along with modest reduction in gun weight due to the barrel change, though the gun itself was already lighter than the MG 151, but slower firing and, of course couldn't be synchronized with a propeller: it was also drum-fed, but the Japanese adapted it to belt-feed)
The Japanese Navy did actually experiment with a 14 mm Oerlikon derivative, not a simple re-chambering adapting existing case with similar bolt-face and such, but a scaled-down weapon (probably using the more efficient straight-walled cases) which was eventually passed over in favor of a browning short-recoil operated gun firing 13.2x99mm. (appealing for synchronization and commonality with their Hotchkiss AA guns)
Additionally, on velocity and barrel wear: this is most severe with relatively heavy for caliber bullets and high pressures and temperatures and is reduced in the cases using light-for-caliber projectiles and lower pressures. Lower bullet weight means lower sectional density and poorer ballistic coefficient, so that's a problem, though BC is improved with longer and more streamlined shapes (an extreme example would be the Spanish 7.92x40mm CETME bullet) so that's always an option provided you have the overall length sufficient for long bullets. (in the case of cannon shells, with the thin-walled, drawn-steel German mine shell type projectiles, density was already reduced a good deal, though ballistic shape was initially quite poor, it was later improved in more complex drawn/formed shapes with a spitzer profiles and boat tail in some MK 108 loadings and probably could have been optimized earlier in 20 mm by reducing the shell capacity somewhat and implementing a long, streamlined, pointed nose cap, while still having much higher capcity than normal bored-steel shells and lower weight + higher muzzle velocity as well)
Additionally, there was at least a private effort to neck .50 BMG up to 16 mm, sufficient for reasonable explosive shells, and Lockheed seems to have been interested in it as a potential P-38 armament, but the weapon was rejected by the Army and details are basically nonexistent.
0.50 Browning MG and it's descendnats for 'other' air forces?
I will say .50 BMG could've been necked out further than that, to at least .70 cal or 18 mm without modifying the taper and shoulder, and .75 cal or 19 mm if the taper was removed for parallel walls and very little or no bottleneck. (20 mm would have required an increased base and rebated rim, if the same case-head and bolt face were used, but perhaps not rebated enough to be useful in API Blowback systems: though retaining that bolt-face and going a bit larger, sufficient for API blowback might have been appealing for manufacturing logistics, especially if an M2 browning derivative was intended alongside competing Oerlikon-derived developments with a common cartridge used) The straight-walled 19 mm cartridge would also have been an interesting proposal for a straight-blowback weapon, though likely a bit heavier than the API equivalent (double the bolt mass is required) and rate-of fire would be reduced, but the design would be simpler and could be made closed-bolt and synchronized. (or open-bolt but with a .303 Browning or MG 17 style synchronization system provided with independent firing pin control and delay provided after the bolt closes until the prop is clear, then striking the primer; this also could have been used on the Hispano, but wasn't, but could not be used on API blowback weapons as the primer is struck as the bolt is still sliding forward and cannot be delayed at all; meanwhile, delayed-blowback systems would also work fine and would be lighter and faster firing than straight-blowback, but still be well-matched to straight-walled cases or cartridges with low bolt-thrust in general: also why the schwarzlose machine gun improved somewhat when switching from 8x50 mannlicher to 8x57 S. Patrone Mauser in Czech use: 8mm Mauser has a lower bolt-thrust, particularly in the light-ball load the Czechs continued using and the Germans had been using prior to the mid 1930s)