Advertisement
Guest User

The program that ate itself

a guest
May 10th, 2015
640
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 1.58 KB | None | 0 0
  1. Relatus had a curious history. Sponsored by Hayward Alker, and taking advantage of the phenomenal social
  2. capital of the Artificial Intelligence Laboratory at MIT, it was probably the most sophisticated effort to come out of
  3. the AI/IR enterprise. However, it also had all of the weaknesses that eventually led to the “AI Winter” that followed
  4. the enthusiasm for AI in the mid-1980s. The system ran on specialized hardware—“LISP machines”—and while
  5. great promises were made for its potential—in particular, we were repeatedly reminded, it was going to work vastly
  6. better than anything that we lowly scum outside of MIT might even contemplate writing—it never got beyond the
  7. stage of a working prototype.
  8.  
  9. Relatus quietly disappeared sometime in the early 1990s. The story told to me, albeit without a lot of detail, was
  10. that the system involved an incredibly complex data base that incorporated a lot of “self-modifying code,” a typical
  11. characteristic of LISP systems, where the program code was treated as just another form of data. A programming
  12. error apparently caused the system to completely destroy itself, and for reasons that aren’t clear, it could not be
  13. restored from backups. [A programmer failing to make regular back-ups?—nah...] The Cambridge-based “LISP
  14. machine” industry was collapsing at the same time because Moore’s Law quickly gave $5000 off-the-shelf personal
  15. computers the same power as dedicated $50,000 LISP workstations, and Relatus was never rebuilt in a portable form
  16.  
  17. from http://eventdata.parusanalytics.com/utilities.dir/KEDS.History.0611.pdf
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement