Design Goals
The goal of the FreeDOS Project is to create another implementation of MS-DOS. DOS is a popular system, and there is plenty of PC hardware already available that is ready to support it. Microsoft will not develop DOS forever, and one cannot count on third-party commercial entities to continue DOS. I feel it is then up to those on the Internet to develop their own DOS (FreeDOS) and I feel there is a lot of support for this type of project.
We intend to write a free replacement for MS-DOS; therefore, our first goal must be to achieve MS-DOS compatibility. We have chosen to become compatible with MS-DOS 3.3, because it is our opinion that MS-DOS 4.x, 5.x, and 6.x are really just improvements upon MS-DOS 3.3. Expanded memory support was improved, and some other things fixed, but on the whole the underlying DOS kernel didn't change that much.
The MS-DOS 3.3 compatibility extends only to the FreeDOS kernel. FreeDOS programs should be compatible with MS-DOS 6.22, because those are the features that users will be most familiar with.
FreeDOS should not be targetted towards certain users. That is, the end product should be something that programmers and non-programmers may both enjoy and find useful.
CPU's To Be Supported
The FreeDOS kernel and all FreeDOS programs (utilities, drivers, compilers, etc.) should run on all levels of PC hardware. FreeDOS should operate on processors as low as the 8088 (XT) with as little memory as 640k.
I do not think there is a problem to distribute optimized versions of the FreeDOS or FreeDOS programs for the 80386 and above. Any FreeDOS program may choose to take advantage of a '386 or other enhancement if detected, but the program should still run on low-end hardware such as the XT.
Exceptions to the XT rule are programs like EMM386, which were originally specified by MS-DOS to run on a '386 or better. Only if a program that was originally distributed with MS-DOS, that specifically says it will run on a certain PC architecture, may be written to support only that PC architecture.