Despite being more powerful, the TI-68k series has always been far less popular than the TI-Z80 series, due to the CAS, forbidden in many standardized tests worldwide, and the higher price tags (even before the Nspire existed). The TI-68k series got fairly popular in France, thanks to the lack of such silly bans on CAS.
Though there have been a number of interesting programs since then, the TI-68k native code programming community saw a massive, final plunge in 2006-2007. It was therefore unrelated to advent of the Nspire (at that time), which became programmable through native code for the general public only in late February 2010.
The dive of the TI-68k native code programming community can be attributed to a number of factors, let's mention at least four of them:
* as I wrote in another topic, evolution of the library and documentation were indeed low because the maintainer actively or passively refused many contributions that were deemed useful by users. Until 2008, there was still some work in other areas. GCC4TI picked up where TIGCC left, fixing bugs, processing some of that backlog of contributions (well, at least, those which were published - we know that some contributions have been lost forever), performing new optimizations in the library, and highly improving the maintainability and portability of the very badly-behaved build system inherited from TIGCC. Little gets done on GCC4TI nowadays, due to the extremely low number of users, but at least, GCC4TI can be built straight from Git, while building TIGCC is pretty hard;
* constant criticism against many projects that could have created more activity, and general disrespect, combined to overly lax moderation (or unfair moderation, when the main troublemaker was the so-called moderator), produced horrible mood on what used to be the main message boards of the TI-68k community, which took a toll on the community. The alternative languages for the TI-68k series, e.g. NewProg (which is close enough in spirit to Axe and predates it by years), or the GTC alternative toolchain, were constantly dissed, or even saw their progress actively hampered through non-technical matters, by the TIGCC maintainer;
* over the years, old-timers of the community moved on to other platforms / interests, until (though it was possibly rather 2008 than 2006-2007 ?) very few old-timers remained really productive;
* newcomers have rarefied due to technical advances, which helped paint calculators for the overpriced pieces of obsolete software / hardware that they are (calculators lost their "first programming platform in students' hands" aspect), and also due to changes in society (there has been a real drop in new graduates in scientific matters - after all, scientific matters are far less valued, though indispensable in the long run, to marketing and finance, so why would people choose lower-wages jobs ?). The Nspire's low programmability and general orientation towards consuming knowledge than producing code, reduces the number of persons which discover programming through the calculator.
In Nspire times (2010+), besides the aforementioned reasons, most of which have remained valid, the TI-68k community didn't pick back up because of TI actively pushing the Nspire instead, through insane price tags and completely dropping support for the TI-68k series (since 2003 for the 89 & 92+, 2005 for the V200 & 89T).
Also, I'd like to mention something about the ticalc.org POTY possibility, which was mentioned by pimathbrainiac above. It's true that it's easier to get a ticalc.org POTY nowadays, especially in that category, but it's not a good thing, or not necessarily because nothing else gets done.
ticalc.org has lost a lot of momentum over the years, due to a chicken-and-egg problem. Reduced newsing activity (despite the addition of a staff member), and upload + update forms that are not on par with those of other sites (I'm being lazy to upload TILP II 1.17 and associated programs due to exactly that...), produce fewer file uploads to ticalc.org, and lower newsing activity...
* in 2010, I received an unexpected POTY, because one of my creations was the only one mentioned in that category over the whole year, though definitely not the only possible one (Benjamin Moody's titools sprang to my mind at the time), and probably not even the most interesting or (potentially) useful one. While novel, and not even that time-consuming, tiosmod+amspatch have seen little use in the wild, for legal reasons: we can't redistribute ready-made modified OS images, so users have to perform manual operations (compiling, or downloading binary patchers and applying the provided binary patches). Anyhow, I didn't spend time on this old idea of mine until 2010, leading to a far too belated tiosmod+amspatch release.
* in the Nspire category, the 2012 POTY vote was significantly truncated by the fact that like half of the important programs of the year were not uploaded to ticalc.org in the first place, or for those which were, no news was posted about them on ticalc.org (despite being featured on Omnimaga, TI-Planet and/or Cemetech).
Oh, and I missed something on the previous page:
But native-code programs can't be larger than 24K; that's a bummer.
The artificial size limitation on ASM programs was added by the AMS 2.xx series in 1999, at first 8 KB, and later raised to 24 KB. However, it's so weak a "protection" that it was immediately defeated, before the 21st century
"ASAP or Exec string too long" and "Invalid Program Reference" have always been useless in practice, neither has prevented large ASM programs or math ASM programs in any kind of significant way. Both are killed by all "kernels", as well as tiosmod+amspatch.
On the TI-Z80 series, until recent discoveries, ASM programs were also limited to 8 KB.
