First of all, your hard data and facts. All the game programming gems sold fairly well, each ranking around the 200k mark on Amazon bestsellers list.
- Gems1: 198,126
- Gems2: 209,933
- Gems6: 200,574
- Gems8: 202,163
So based on that data I wouldn't say lack of sales is a good reason to stop.
But what about the content. "Game Programming Gems" was like a really fat magazine that came out every couple of years and costed $60-$80. After they covered the basic topics that everyone can understand in the first couple of books (quaternions & slerp, basic AI architecture), the articles seemed to get more and more advanced. Just look at a couple of title from the TOC of GPG 1:
Gems 1:
- 1.1 Object-Oriented Programming
- 2.0 Predictable Random Numbers
- 3.1 A Finite-State Machine Class
- 3.2 Game Trees
- 3.3 A* for path finding
Gems 3:
- 3.4 Terrain Analysis in an RTS - the hidden giant
- 4.4 Fast and Simple Occlusion culling
- 6.6 Stochastic Synthesis of Complex sounds
Gems 4:
- 2.4 Nonuniform Sprites
- 3.6 Interactive Water Surfaces
- 5.9 Fast Sepia Tone Conversion
- 5.12 Hardware skinning with quaternions
Gems 8:
- 1.3 Multi-Resolution Deferred Shading
- 2.5 Improved Numerical Integration with Analytical Techniques
- 3.5 Applying Control Theory to Game AI and Physics
So you can see, as the series goes on, 2 things happen:
- The topics get more advanced, which means the articles take more effort to write, and the book will no longer appeals as much to total beginners
- The topics get more narrow and specific, because they are so advanced, which means you'd probably find a smaller proportion of the later books in the series useful, unless you have really wide interests (and abilities).
So although I think that shouldn't PREVENT additional installments into the series, I do think that in a way the series raises its own bar higher and higher all the time (both barrier to write an article, and barrier to read/understand an article).
In conclusion, I dunno. You'd have to ask Mark DeLoura.