Recently, I have noticed that there seems to be an increasing number of projects for the CE (especially due to the hype about the new colors) and a similarly decreasing number for the 83, 83+, etc. What do you think about this? Will the CE replace the 84+ or do these kind of popularity shifts happen often and not do much?

Definitely not.
Even if all world starts using CEs, that only means there will be cheap monochrome ones on sale and I will buy one.
I mean it!

TI has really been pushing the CE to replace the aging hardware of the TI-84 Plus (a 2004 remake of a 1998 remake of a 1996 remake of 1993 hardware+software), and the previous attempt with the now-forgotten TI-84 Plus CSE failed. This time around, they actually did a good job at making a programming-friendly color calculator largely compatible with the TI-83 Plus. The new, more powerful hardware means developers are less restricted, and they can now write less efficient more advanced code; this is inviting to developers and stole the limelight from the monochrome calculators and CSE.

It's a real shame attention isn't where it belongs: the TI-82. /s

I think that B/W and colour calculators will live together for a very very long time.

Colour screens with backlight consume more power, and sometimes it is good to have 5 months of power with the same batteries.

Although, programming a monochrome calculator has an special flavour.

P.D: You can tell this to my classmates, they don't know that colour calculators exist.

I wouldn't go so far as to say that the TI-84 Plus C Silver Edition failed. I believe that the point of that (rushed) project, based on my own assumptions and a few things I've heard, was to determine if people actually wanted a color model in the TI-84 Plus family. Because people actually bought it enthusiastically instead of sticking with only the TI-Nspire models, I believe that they then decided to put more engineering effort and money into making an actual "updated" color TI-84 Plus family calculator.

I'd say it failed. TI hardly advertises it, the OS is still unavailable from TI, and it will likely never receive any updates. It was effectively end-of-life before production ended, and the CE replaced all of its functionality except CBL/CBL2 operation and CSE-specific assembly code. The CE is also (almost certainly) cheaper to produce, and thus more favorable for TI/Kinpo to produce.

If your metric is "success and longevity", yes, it failed, but I don't think that was the reason TI made it. Other than being a product, I believe its primary purpose was to determine if people actually wanted a color TI-84 Plus. Since they did, they made the TI-84 Plus CE.

Indeed; as a product its performance was unacceptable and the market responded like such. The concept was proven to be good, but the execution was too poor to consider the CSE a success.

Seeing how popular analysis of how people use certain products became in the last years (Google Analytics used in literally half of all software, polls, tracking and so on), it raises the question, wether TI couldn't simply make a poll and ask people wether they want a color calculator?
Because developing a new calculator just for testing purposes clearly is more of an effort, even if it yields better results.

On second thought, did the 2011 Nspire CX have similar expectations, to see if people wanted a color screen on a calculator; or was it merely an attempt to justify a price division beyond software functionality?

I'd say it was the obvious evolution for the Nspire series.
Casio did the same thing with the Prizm.

Can't get over the fact that the color nspire is quite buggy and somehow manages to lag a lot more than any ti-84 regardless of it's superior hardware. I don't call that an evolution in my book (I've only used a monochrome nspire once and it didn't seem buggy and glitchy) I still kind of feel the same way about the CE although it's a lot better. For some reason, the CE feels a little clumsy (random crashes, although that might be because of sketchy asm code) and some strange (unreproduceable?) bugs that leave if the calculator is left to sit for a day or two but not nearly as much as the nspire. I guess the more complicated the hardware gets, the harder it is for engineers to create software for it.
Anyways, I'm getting a little far off of the topic, I too have noticed the CE seems to be taking over the forums but also in real life, last semester I started seeing random students popping up with CEs (not just the geeky ones), it seems like when people show up to buy a graphing calculator, they see the bulky monochrome 84 + and then the flashy CE in a similar price range and just go for the fancier one

IMHO the monochrome TI scene is now suffering the same fate as the TI-82 and 83 did when the TI-84 Plus became popular. The same thing happened to every last-gen calculator model and will happen to the TI-84 Plus CE if it gets replaced in a few years. There will still be a lot of people making educational programs for monochrome models for the next years or so, but much fewer game programmers.

Also the monochrome TI Z80 models still being sold suffers from the fact that the only ones that supports ASM and Axe are sold exclusively in North America. The TI-83 Plus, TI-84 Plus and TI-84 Plus Silver Edition in France were replaced by the TI-82 Advanced and in Europe by the TI-84 Plus-T, both of which lacks ASM support by default.

Unless BrandonW ever decides to release his hack to run ASM programs on the TI-82 Advanced and adapt it for the TI-84 Plus-T, monochrome Z80 ASM development will most likely eventually become solely a North American thing.

Addressing the rise in development for CE, a huge factor is the support of C. It's a language many people already know and, if not, much easier to learn than assembly. People hated the Lua support for Nspire since it was horribly slow, so this is TI's redress for that. Consider as well the improvements in clock speed, memory, and flash.

On top of that, people have realized that $100 does not justify a twenty-year-old monochrome calculator that should only cost $30 parts and labor (you could buy an Android tablet with $100!), so TI's natural response is making a color calculator that looks like it's keeping up with the times and thus easier to market. TI is banking on people's tendencies to buy "newer" things because they appear to be worthwhile investments.

The 68k calcs had support for C and extensive library availability, yet that was not enough to let the platform gain popularity. The CAS felt 'cheaty' to educators, and they lacked test acceptance that the Z80 calculators had. The 68ks also had a powerful BASIC that other calculators should have implemented. Even if a platform is capable of being popular for development, it's hardly a factor in how much code will get written, since 99.5% of users don't consider programmability a purchasing factor.

Wow, there are a lot of opinions on this! It seems to me now that the calcs will live on as long as there is a market for them. Who knows.

It's getting harder for TI to source parts for the monochrome calcs (everything is moving towards low-power RISC devices / SoCs), so to keep production volume up they can (and will eventually have to) use more readily available parts. When these devices were new in the days before Windows 95, their performance characteristics (graphing time, memory, display density/contrast, lack of backlight, etc.) were considered good. All the way up to the CE, we failed to see improvement in these except for the CSE's color screen, which came to the detriment of other factors. By this point, even the cell phones of yesteryear were more powerful than the 15 MHz, 8-bit processor seen in current calculators, and the slow performance resulting from this was simply an expectation of graphing calculators. We definitely needed something new; the Nspire tried, but (among other things) its complexity caused it to fail. The CE came along, and brought usable performance to a familiar platform compatible with textbooks. It's just about all-around better; it's significantly thinner than the AAA-powered devices before it, uses readily-available components in its construction, and retains most of the functionality of the TI-83 Plus. It's marketed as something new, a replacement of old standards -- thus, I do believe it is intended to end the TI-83 Plus' hold on the graphing calculator market, and there is evidence showing it killed at least the TI-84 Plus. So while the monochrome calculators may live on, manufacturing probably won't.