There are many references on Google to versions of AtariLink that seemed to be out in the wild. I don't know for sure if it was free or commercial, or if this 2.0 prototype version ever made it to a release.
Very nice! I would guess these must be the released version. Would be an interesting project for someone with more time than I have to do some comparisons of code between the prototype and the release.
I've had some luck getting a fresh copy of the work disk after a number of tries.
Check out the crazy stuff going on between 1:24 and 1:30 in the video ... Looks like some self-modifying code, maybe deleting some lines no longer needed and doing some other stuff.
This is an old thread, but I had to jump in here to say that's almost certainly what is going on.
If an Atari 8-bit BBS in BASIC was running tight on memory (which one of the screenshots said it was), they'd resort to a number of tricks to gain some of it back. One of the things done early on would be to replace the most commonly used numbers (like 0-10, 20, 50, 100, etc) with variables C0, C1, C2 ... C100. After some initial overhead, even if the variables were used just a few times, it ended up saving space. (And more so with each subsequent use.) Another method (seen here) was to clear out any initialization code that wouldn't need to run a second time. But if you had to restart the BBS, you'd just load the initial BASIC program once again.
Usually clearing out lines of code is done with an ENTER command with a locally stored file. What is unusual about the method that they used was that they seemed to read in and display a text file, reposition the cursor at the top of the screen, break to immediate mode, and then force carriage returns to step through the modified lines. This was repeated a for a second screenful with a new set of commands, finally ending in the code which store the new number of bytes free, and then likely jumping into their main routine which waits for a call.
Not unrelated is why the deleted lines were 0, then in the 32000+ range. Line 0 would have been a jump to 32000, which is where most of the initialization could would have been located. Why? Because the way the interpreter worked, the initial lines of code in Atari BASIC ran the fastest, and the last lines of code ran the slowest. (The more lines of code that you had, the more true this was.) Someone else will probably chime in with the details, but if memory serves me, Atari BASIC would do a linear search for a line number. The first lines ran faster because it literally could find them faster. So the low numbered lines are where you'd find the simple input/output and string handling routines in BBS software.
This is some of the trivia that you learn but never forget after modifying a number of different pieces of BBS software (and then taking those lessons and coding your own). I'm going to enjoy looking at the software. Thanks for the disks!