Barbara Liskov Wins Turing Award 187
jonniee writes "MIT Professor Barbara Liskov has been granted the ACM's Turing Award. Liskov, the first US woman to earn a PhD in computer science, was recognized for helping make software more reliable, consistent and resistant to errors and hacking. She is only the second woman to receive the honor, which carries a $250,000 purse and is often described as the 'Nobel Prize in computing.'"
Re:Turing test (Score:1, Interesting)
making software more reliable? (Score:4, Interesting)
Software is ALWAYS reliable. It is the code that people write that sucks.
I don't know how many people come from the "old school" of programming, but when I started, we didn't have all these libraries to link to. When we wanted a function to happen, we wrote it. And when we wrote it, we checked for overflows, underflows, error status and illegal input. We didn't rely on what few functions that already existed.
Most fatal program flaws are ridiculously easy to prevent, but bad programming habits prevail and short of creating some human language interpreter that writes code as it should be written, nothing will replace lazy programmers who trust library functions too much. And yes, I know about deadlines and not having time to waste and all that stuff. But there is something most people are also missing -- pride! I know that when I do something, I am putting my name on it whether it is directly or otherwise. And if my name gets associated with something, I make damned sure that it works and is of good quality. With the stuff that goes out these days (especially SQL injection?! PLEASE! What could be more fundamental than screening out acquired text data for illegal characters and lengths?!) it is clear that pride in one's own work is not something that commonly exists.
For those of you out there who agree with me, it probably doesn't apply to you. For those that disagree, tell me why? Why is a programming error FIXABLE but not PREVENTABLE?
LSP it's not a guideline, it's a rule. (Score:5, Interesting)
She deserves recognition for the vast number of latent defects she has effectively removed from the worlds software with the LSP alone, I'm glad she got the award.
Re:making software more reliable? (Score:4, Interesting)
No, quantum mechanics is reliable. It defines physical uncertainties in a robust way. Electrons suffer from crappy positional and momentum certainty.
Re:making software more reliable? (Score:3, Interesting)
Functions? Back when I started, we didn't have functions. We had jump instructions.
When I first learned to program as a kid, I taught myself how to write pseudo-functions using goto. I looked back at those programs a few years later, and they were completely unreadable. Now, my wife does a little programming on the side (we're both researchers) and she loves goto. I keep trying to tell her NOT TO USE GOTO, but she never listens. It's painful to read her code. I think we might need counseling for this...
More women in the old days (Score:5, Interesting)
Apparently there were far more women in computing in "the old days". The dominance of the male geeks is a relatively recent phenomenon.
Re:LSP it's not a guideline, it's a rule. (Score:3, Interesting)
Proof, please; you are contesting an award-winning theory, and I for one side with prevailing theory until further evidence is provided.
Re:LSP it's not a guideline, it's a rule. (Score:3, Interesting)
What are you talking about? It's not a theory, it's a definition. One specific definition out of several, I would imagine. Moreover, nothing in it says "OBEY ME OR YOU'RE BUGGED"; it's a very good guideline for sensible program design, but it has nothing to do with the completely independent definition of a bug in the general sense. Of course, you can always *make* the LSP part of your design criteria so that violations of it do in fact constitute bugs for your project, but that's a different matter.
If for each object o1 of type S there is an object o2 of type T such that for all programs P defined in terms of T, the behavior of P is unchanged when o1 is substituted for o2 then S is a subtype of T
If you write code to work with a Polygon reference but it fails to work with a Triangle reference, and Triangle is a subclass of Polygon, that does not indicate that you have a bug in your code. It merely indicates that Triangle, despite being a subclass, does not constitute a subtype under this definition.
If that's not good enough for you, notice that the statement is an if, not an iff (if and only if). Therefore, that sentence alone does nothing to rule out what can be a subtype of T.
Re:1968 (Score:2, Interesting)
> you're either trying to be funny (but got modded +3 insightful), or are seriously trying to imply that a woman who's good at C.S. is as much of a freak as a pregnant man.
Try neither. I had no emoticon, and I had no implications -- I just asked a simple question, in order to why find out where this assumption is coming from.
I have seen this slippery-slope type of Political Censorship before and all it does is lead to reverse discrimination. Replace Women with Ethnic / Religion / X of your choice.
This completely focuses on the wrong problem by making a problem where one doesn't exist. The gender of a Computer Scientist doesn't fucking matter -- the only things that do are...
a) Are they competent? How well do they know their doman? How well can they solve problems?
b) How professional are they when interacting with
i) colleagues?
ii) the layman?
A better question to ask is "How effective are we teaching computer scientists?"
I still have never seen a good answer to this question.
Re:Turing test (Score:3, Interesting)
While there's no doubting her accomplishments, I will say that my enjoyment of 6.170 was in spite of her.
Not surprised.
I had no use for her in 1978. She actively assisted in flunking a good friend of mine out of the PhD program. She turned down a thesis idea I had, called it "totally the wrong direction" -- and three years later a guy got a PhD and an award with the same idea at Waterloo.
Maybe she mellowed with age, but given your comment, I guess not.
Re:1968 (Score:3, Interesting)
Better rhetorical questions:
* Do you complain we need more male nurses?
* Do you complain we need more male teachers?
* Do you complain we need more female garbage collectors?
Gender equality is not the same as having a 50/50 male/female split in every field.
Re:Coincidentally (Score:3, Interesting)
I survived 8 years as a OO programmer without hearing this term. Even then, only as a question on a recruitment agency test.
So either I went to the wrong university, and consistently the wrong employers, or it's one of those self-evident principles that just didn't have a name before Barbara turned up.
In context of history of languages (Score:3, Interesting)
CLU drew on the lessons learned with both Alphard and Vers. Alphard was from CMU and written by Wulf and Shaw, Wulf also writing the famous BLISS. Vers was made by Jay Early, whose parser was hugely important in all the compilers of the time. THe language itself (and its V-graphs) was heavily influenced by the Mem-theory of Anatol Holt (who was on the Alogl Committee and was a principle in designing the astonishing GP and GPX systems for the UNIVAC - first languages to explicitly feature ADTs per se. That became ACT, the adaptable programming system for the Army's Fieldata portable computers (portable in a completely different sense to the modern usage. He also hated Unicode, but that was a rival programming system back then. So reading the reports at the time can be misleading - "don't use Unicode on Portable Computers!"). Holt's ideas permeate computing, the notion of making any system of data representation as abstract as possible goes back to him.
CLU was written using MDL (pronounced muddle) which was a protoreplacement for LISP which featured ADTs. MDL was cowritten by Sussmann of LISP fame as a basis for PLANNER which became Scheme, and perhaps more geekly interesting is that is was also used for writing ZIL (and if you don't know about ZIL, you shouldn't be reading Slashdot)
CLU evolved into Argus, but the ideas were also used in the Theta programming system for the Thor OO database, and was also in PolyJ which was (as it suggests) a Polymorphic Java
Another fascinating development of the CLU ideas the SPIL system that Liskov co-wrote at the USAF-sponsored MITRE corp, which was in turn used for writing the VENUS operating system
Liskov has pioneered the notion of abstraction per se in language design for 40 years, and this generics-based approached is now taken for granted. She fully deserved the award for her insights as well as for her determination in fighting the reductionism represented by previous recipients (eg Dijkstra) although opposed by others (eg Iverson)
I have extracts for reports for all language of these on the HOPL website HOPL.murdoch.edu.au (too many URLs to paste in individually). Find CLU http://hopl.murdoch.edu.au/showlanguage.prx?exp=637&name=CLU [murdoch.edu.au] and follow the genealogy links. And if you haven't yet seen my 4000-strong programming-language family tree it is worth printing out for wallpaper if you have an A1 plotter.