1. The language should easy to be accepted. To make a language easily acceptable, it has to be available on a wide variety of hardware, it should be easy to learn, it's implementation should be simple for it to spread and lastly, it should be similar to existing language.
2. Successful languages require minimal ore modest computer resources. It should carefully choose the problems to through innovation and maintain the computing speed achieved by hardware development.
3. Successful language should provide a simple model to access its performance. This will help programmer write more efficient program.
4. Successful language should have mathematical simplicity.

In his paper Gabriel missed one equally important point that directs the creation and development of programming languages. That is the social need. By looking at the history of programming language [2], every language is created from a need. As long as such a need still exists and there is no better substitution, the language prevails. The development of Lisp gives a good clue on how the "need" theory and Gabriel's "simplicity" theory play a role in a programming language's fate.

By using Gabriel's theory, Lisp is considered to be in ill health. Indeed, showing on his statistical data, Lisp code is only a fractional of all the code that has bee written.

But, Gabriel's use of his static data to measure how popular a programming language is and it's chance of survival is problematic. The popularity of a language should be measured not only by how many people are using it and how many lines of code are written by it, but also by how many people are using it to accomplish the kind of task which this language is designed for.

Lisp is not designed for general programming, it is born from a desire for an algebraic list-processing language for artificial intelligent work [1]. Since its creation in the early 1960's, LISP has remained major language of choice for implementing state-of-the-art programs that simulate human intelligence:

• LISP is a recursive language that makes it an ideal formalism for describing complex concepts.
• LISP is a list processing language that can effectively simulate a wide range of real-world problems.
• LISP uses dynamic memory management to automatically recycle and reallocate memory resources according to the changing demands of programs as they run.
• LISP is a data-driven language that allows data structures to be created and then executed as a program.
• Lisp is still the best prototyping language.

References:

[1]  R. P. Gabriel.  The end of history and the last programming language.  Journal of Object-Oriented Programming,  6(4):90-94, July 1993.

[2] Dennis M. Richie.  The development of the C language.  Second History of Programming Languages conference, Cambridge, Mass., April 1993

[3]Richard L. Wexelblat.  History of programming languages.  New York : Academic Press, 1981.
  developers exist and grow, Lisp language will keep to survive and develop.