Are there Prolog JOBS out there?
Ahmed Bouzid
Dear Prolog people:
I have been programming in Prolog for the past six years--all
kinds of flavors, versions, etc.--but all in university setting
and research oriented. Question: is there any chance at all of
finding a job in the "real world" programming in Prolog, or
is that chance so slim that it's a fat chance?
Later,
Ahmed
Michael Covington
: Dear Prolog people:
The thing to do is hook up with Quintus, Arity, and other Prolog vendors
and find out who is doing practical software development in Prolog.
(By now, there's a significant amount; there's even rumored to be some
Prolog in the Windows NT installation routines.) My students have had
no trouble finding jobs in the software industry.
--
< Michael A. Covington, Assc Rsch Scientist, Artificial Intelligence Center >
< The University of Georgia, Athens, GA 30602-7415 USA mcov...@ai.uga.edu >
< Unless specifically indicated, I am not speaking for the University. > <><
> "To 'see through' all things is the same as not to see them." - C. S. Lewis <
Ramesh Kumar Rayudu
>The thing to do is hook up with Quintus, Arity, and other Prolog vendors
>and find out who is doing practical software development in Prolog.
>(By now, there's a significant amount; there's even rumored to be some
>Prolog in the Windows NT installation routines.) My students have had
>no trouble finding jobs in the software industry.
Recently, I read "Dr. Dobbs" magazine (june/july issue I think) which states
that Windows NT networking installation is written in prolog (SB=Prolog).
Infact, they also included a small prolog interpreter (SB's) into the
software itself.
>< Michael A. Covington, Assc Rsch Scientist, Artificial
Intelligence Center >>< The University of Georgia, Athens, GA 30602-7415 USA
mcov...@ai.uga.edu >>< Unless specifically indicated, I am not speaking for
the University. > <><>> "To 'see through' all things is the same as not to
see them." - C. S. Lewis <
Ramesh.
Ken Bibb
>(By now, there's a significant amount; there's even rumored to be some
>Prolog in the Windows NT installation routines.)
It's no longer just a rumor :) In the August 1994 Dr Dobb's Journal
(which I purchased because of the words "Prolog" and "Lisp" on the
front cover) on p. 80 there is an article "Small Prolog and Windows
NT Networking" by David Hovel. Some excerpts:
Windows NT networking installation and configuration is
controlled by the file NCPA.CPL, which the user sees as the
Networks icon within the Control Panel's main window. The
bulk of this DLL is written in C++, but it also contains a
simplified Prolog interpreter known as "Small Prolog"...
....
Along with its C++ code and SProlog interpreter, the
NCPA.CPL file has a 700-line SProlog program eembedded
into it as a textual Windows "resource." The program
is the actual network-configuration algorithm...
Mark Kantrowitz
similar mailing lists for lisp-jobs and ai-jobs.
--mark
Subject: [1-15] Prolog Job Postings
The PROLOG-JOBS mailing list exists to help programmers find Prolog
programming positions, and to help companies with Prolog programming
positions find capable Prolog programmers. (Prolog here means Prolog-like
languages, including logic programming languages.)
Material appropriate for the list includes Prolog job announcements and
should be sent to ai+prol...@cs.cmu.edu. Resumes should not be sent to
the list.
To subscribe, send a message to ai+q...@cs.cmu.edu with
subscribe prolog-jobs <First Name> <Last Name>, <Affiliation/Organization>
in the message body.
For help on using the query server, send mail to ai+q...@cs.cmu.edu with
help
in the message body.
If you have any other questions, please send them to a...@cs.cmu.edu
David Cohen
I hope you will forgive me for taking so much space to answer your
simple question about the availability of Prolog jobs. I hope that this
reply of mine will help guide you and others in your position.
Introduction: Some Facts
------------------------
First off, before I offer you my own, admittedly subjective experiences and
the subjective opinions they engendered, let me present you with some facts:
Examine the want ads in the newspaper and you will find no Prolog jobs.
Examine the postings in misc.jobs.offered you will find no Prolog jobs.
Examine the postings in misc.jobs.contract you will find no Prolog jobs.
Call around to various job recruiters and you will find no Prolog jobs.
In fact, were you to examine all these sources over the last 5 years as I have
done, you will still find Prolog jobs as scarce as hens teeth.
As far as calling the Prolog companies, Arity's bulletin board is fairly dead,
as usual. Not likely you'll find any leads there. The staff of Arity in my
experience was not able to point me to anything and they themselves don't often
have openings, since they have been a small company. On the other hand, they
have moved in the last year or two, just don't get your hopes up. AAIS, the Mac
folks, are not going to be developing their Prolog any more according to
postings in this newsgroup. Even Quintus, the flagship Prolog company, has not
been pushing Prolog for a long time, but now they have even stopped
advertising. I called them recently to check up with acquaintances there.
Everyone I named
including long-time star employees such as Tim Lindholm were gone. The
erstwhile president is just that--erstwhile. They have shrunk considerably and
are pegging their hopes on applications that they have developed such as
Customer-Q, something along the lines of a help desk system. They hope to get
free
of Intergraph again one of these days but finding the money to do it isn't
easy. In the past I called them to see about consulting opportunities, but
nothing came of it. Haven't heard much out of PDC, the folks who developed, and
now control, Turbo Prolog, but they are around in Atlanta I believe and might
be worth talking to, even if their Prolog isn't *really* Prolog. These days, a
job is a job is a job.
Where Prolog Seems to Be Heading--My Experience
-----------------------------------------------
I came out to LA to work for BIM Systems Inc, the US subsidiary of BIM Systems
in Belgium after finding that job hunting in recession-bound Boston was very
hard and the preponderance of Prolog on the resume just made it harder. (BIM is
the European competitor to Quintus and for the record I am the d...@bim.com
mentioned in the Prolog FAQ acknowledgements for no good reason). Anyway in my
position as the US company's point man for Prolog, I had the task of calling
BIM's customers to see how they were doing, given that BIM had
transferred the responsibility of supporting customers from a free-lance
individual in San Diego to the LA-based BIM subsidiary. I spoke to many
customers who had used ProLog by BIM, some of them even commercial users. I
also called many of the attendees of ICLP '90 who had expressed an interest in
BIM's Prolog. The response that I got back was primarily that either the Prolog
was gathering dust or that whatever system they had written in Prolog was now
ported to C.
I made a point of calling others whom I knew had used Prolog. For example,
Prolog was in use at the FBI for a rather large project that I had known about.
I was told that almost all of it had been ported to C.
Perhaps you saw a story in this newsgroup by a guy on Wall Street who had
done a Prolog prototype which like many such prototypes was good enough
to be put into production and began making money hand over fist. Management
though decreed it should be redone for production in C++ since this is what
real programs are written in. They wasted a year of work. But the Prolog
system was so successful that it more than made up for the cost of the
abortive C++ effort. Still they decided to reimplement in C++.
Getting A Prolog Job
---------------------
Now this doesn't mean that no one is using Prolog or whatever, but
most folks either don't know about it or don't care about it or have used
it and were burnt by bad performance in earlier implementations, bad
interconnectivity, arrogant vendors or by their management. When you do
find a Prolog slot, my experience suggests that it will be with small
undercapitalized, perhaps academically-oriented, outfits. This means that
you cannot expect to hold out for a Prolog slot, thinking that you will
make $1000/day when you finally hit it big. Better to just stick with
lottery tickets. They cost less, pay off bigger and you are more likely to win
the jackpot than with Prolog.
Perhaps you will find a half decent job involving Prolog. Congratulations.
Say it lasts and the company doesn't collapse, like my former employer BIM
Systems US did (only the US company--they are OK in Belgium). But then a little
while down the road you want to move. If you haven't developed some solid C,
C++, Smalltalk, telecommunications, client-server, or other commercially
viable skills, you had better be prepared for a long job hunt since employers
are being very picky and are not likely to give you extra points for knowing
an exotic language like Prolog. If you don't know SQL or possibly more
correctly, if you don't know Sybase or Oracle, the fact that you know all
about relational database and set theory from your Prolog experience will not
amount to much.
This isn't to say that Prolog isn't a wonderful language that I have had
a good time learning and using. This isn't to say that logic programming is not
a worthwhile academic endeavor. But if you are expecting to see Prolog
finally have its day, better not hold your breath. The fact that Microsoft has
embedded a bit of Prolog into the bowels of Windows NT is utterly meaningless
in the larger context of the commercial programming world.
Reality Check
-------------
Am I perhaps alone in feeling this way? Am I just a disgruntled ex-Prologer
who sold out to the establishment? Could be, you don't know me after all.
So let me tell you the story of a man who was quite active on
comp.lang.prolog,
whose first name is Eduardo I think and his last name might be Lagauche or
something like that. Anyway, he was very, very interested in Prolog and logic
programming. He put out a newsletter. On *paper* which he sent to intereted
parties, such as myself. He even paid the postage. One day there was a message
from him in this very newsgroup. It was essentially a "goodbye, cruel world"
letter. He couldn't find work and was rather bitter about it. He had no desire
to be involved any more with Prolog from what it sounded like and I cannot
remember ever seeing him again in the newsgroup.
This leads me to my conclusion which is, while things may be somewhat
different
in some other countries such as England, where Prolog is bigger, things here
are very grim if one is looking for a Prolog job. I must say that I am very
displeased to see members of the academic community try to suggest that things
are otherwise, since encouraging people to seek their fortunes in the Prolog
marketplace as it now exists is just short of criminal.
No less than Leon Sterling wrote in his introduction to the second edition of
"The Art of Prolog" that Prolog seems to be doing so much better since the
first edition came out and that this is evidenced by such indicators as the
Prolog 1000, a list of 1000 real Prolog applications. I think someone must be
spiking his tea. As documented above, Prolog is not thriving and the Prolog
1000 to my knowledge has not yet even reached 1000 and of course many of these
applications were done overseas, meaning that they imply nothing about the
American marketplace and even if 1000 such applications were done, this is
about as meaningful as advertisements from software vendors suggesting that
major universities and companies have purchased their products. So what? So
what if 1000 applications exist if no one knows or cares about them except the
denizens of the Prolog ghetto? And beware seeing
the same projects mentioned in articles in magazines written by the same few
people. More evidence of the ghetto.
In short Ahmed, to paraphrase the Rubiyat of Omar Khayyam--this from ruba'i
13:
(my apologies to Omar Khayyam and Edward Fitzgerald)
"Some for the Glories of This World; and some
Sigh for the Prolog Paradise to come;
Ah, take the cash and let the Prolog go
Nor heed the rumble of a distant drum!"
for the distant drum rumbles but always in the distance and usually it is
beat by the same people who have been beating it for years to no avail,
even when they get front page articles in trade publications like
Software Development (formerly Computer Languages).
To sum up, commercially speaking, Prolog is like the Amityville Horror:
"GET OUT, GET OUT!!!!"
__
DC
P.S. Good luck whichever way you turn. E-mail me if you have more questions.
Michael Covington
: "GET OUT, GET OUT!!!!"
Hmmm. Maybe we have different ideas about what a Prolog job is. Our
graduates are having no trouble getting jobs -- but not pure Prolog coding
jobs; more like knowledge engineering and problem analysis with Prolog as
one of several tools. They know Prolog but are not dead set on using
it as their only tool.
I suppose it's rather like looking for a cabinetmaking job vs. looking
for a job operating a radial-arm saw.
David Cohen
in the commercial sphere:
> Hmmm. Maybe we have different ideas about what a Prolog job is. Our
> graduates are having no trouble getting jobs -- but not pure Prolog coding
> jobs; more like knowledge engineering and problem analysis with Prolog as
> one of several tools. They know Prolog but are not dead set on using
> it as their only tool.
> They know Prolog but are not dead set on using
> it as their only tool.
> I suppose it's rather like looking for a cabinetmaking job vs. looking
> for a job operating a radial-arm saw.
No, no, no and no. A thousand times no. The analogy might be appropriate for a
COBOL programmer or somesuch, but certainly not for a Prolog developer, given
the intellectual development that a Prolog person goes through.
Anyway, to the main point: my definition of a Prolog job is a job where the
skills I have developed through "The Practice of Prolog" will put bread on my
table and a roof over my head.
The job might exploit all, some or none of the following skills that might be
expected of a competent Prolog programmer: logic, rule-based programming,
relational database theory and practice, perhaps some SQL, data-driven
programming, rapid prototyping with a high-level environment such as Prolog,
Lisp or Smalltalk, natural language processing, knowledge engineering,
artificial intelligence, and so on.
In addition, having spent time working with Arity's UI building predicates, I
could add experience with event-driven programming, object oriented UI
programming, creating data entry components with data-validation built-in and
so on.
And of course, I had some C as well as exposure to PostScript, Smalltalk and
other things as well.
The companies not doing Prolog did not care. They did not see that someone who
spent time getting into all of these issues is probably able to work
competently in a variety of different areas. Companies interested particularly
in Prolog were virtually without exception small and undercapitalized.
I was not dead set on using Prolog. I was hoping to be able to use better than
average tools and environments to work on more interesting than average
projects. Things did not work out that way. Ironically, the Oracle work I am
now doing, which pays well and regularly, came to me because of the SQL skill I
had developed as part of some Prolog work I had done years ago. I was very
lucky, though, since most managers and certainly all recruiters, would not
likely have responded well to the fact that I had no Oracle, just generic SQL.
To put it plainly, logic, both as an abstract concept and as reified in Prolog,
does not apply to the marketplace.
I invite you to provide counter-examples of serious companies offering serious
jobs paying a serious wage for serious work using Prolog even tangentially.
But it is important to remember, that even if there are a few companies here
and there where the practice of Prolog is respected and rewarded, these are
likely exceptions that prove the rule. As I believe I mentioned in my previous
posting,
even if Microsoft uses a bit of Prolog in the bowels of Windows NT, so what?
If a rap song incorporated a digital sample of Beethoven's Fifth, would I be
justified in suggesting that this demonstrates that classical music has had a
deep influence on Hip Hop?
One more point: it is possible to use Prolog and make money, but only if one
is able to build a consultancy based on providing solutions to problems. In
such a situation, one might use Prolog or whatever one desires as long as
the projects come in on time and on budget. Trouble is, like any consultancy,
this is far from easy to arrange. Furthermore, even if one can do the programs
in Prolog, how easy will it be to arrange for someone else to take them over
in case you get hit by a bus or whatever?
So, like it or not, as a collegue of mine once told me when he saw me
grappling
with Arity bugs all those years ago, "Prolog is cursed". Truer words are
rarely spoken.
__
DC
Fernando Pereira <8414-42463> 0112220
(co...@zurich.ai.mit.edu) gives a detailed account of his observations and
experience with the lack of penetration of Prolog and the lack of "Prolog jobs"
in the US business world. I have no reason to dispute his comments, but I would
like to argue that the main problem with "Prolog jobs" is that "X jobs" where X
is some tool, are an illusory and dangerous notion. Where are the PL/1 or PDP-8
jobs of yesteryear? They may still exist in the limbo of legacy applications, but
they certainly are not growth jobs. What matters in the business world, and even
in the research world, is problem-solving abilities. At any given time, certain
tools dominate any given problem class, but what stays much longer than the
particular tool is the expertise in the problem class, its history and evolution.
An average career lasts 40 years (and increased longevity and fitness are pushing
that number up). What will keep one in rewarding employment is the ability to
keep on top of problems in a particular class, learning and creating new solutions
for them. In so far as logic programming provides useful tools for thought (and it
definitely does), it helps foster the required problem-solving ability and so it
deserves being thaught. But anyone who believes that knowing a particular tool
intimately, be it Prolog or C++, is a foundation for a good career is risking
deep disappointment.
Fernando Pereira
2D-447, AT&T Bell Laboratories
600 Mountain Ave, PO Box 636
Murray Hill, NJ 07974-0636
per...@research.att.com
Matthew Huntbach
: What matters in the business world, and even
: in the research world, is problem-solving abilities. At any given time,
: certain tools dominate any given problem class, but what stays much longer
: than the particular tool is the expertise in the problem class, its history
: and evolution.
Is this "what matters" or "what should matter" in the business world? Sadly,
it does seem to me, at least here in the UK, that most job adverts ask for
skills in specific tools, and because of this students are demanding that
use of those tools is taught in an academic environment even if this is at
the expense of long-term deeper problem-solving skills. There's pressure to
take "it's good for you, but you might not appreciate it" things like logic
programming out of the curriculum for fear that they put off good potential
students.
If business really does agree that what matters is problem-solving abilities,
I wish it would do a bit more to make this clear to the next generation so
that they are motivated to take this sort of thing seriously when academics
want to to it to them.
Matthew Huntbach
Andre Vellino
(Matthew Huntbach) writes:
|> Is this "what matters" or "what should matter" in the business world? Sadly,
|> it does seem to me, at least here in the UK, that most job adverts ask for
|> skills in specific tools, and because of this students are demanding that
|> use of those tools is taught in an academic environment even if this is at
|> the expense of long-term deeper problem-solving skills. There's pressure to
|> take "it's good for you, but you might not appreciate it" things like logic
|> programming out of the curriculum for fear that they put off good potential
|> students.
I think Fernando is right, problem-solving skills are what matters. Job
ads are skills-specific because of the costs involved in training
employees in the use of specific tools. What gets ignored is the greater
long-term costs of hiring people who can't understand what the problem
is and solve the wrong one or can't solve the problem at all.
Speaking of Prolog education in the UK, I found a rather surprising
statement on page 1 of the Draft ISO Standard for Prolog:
"Prolog is taught in more UK university computing
degrees than any other programming language"
(as though this was a good reason for standardizing the language).
---------------------------------------------------------------------
Andre' Vellino, vel...@bnr.ca, (613)763-7514, fax (613)763-4222
Bell-Northern Research, Box 3511, Station C, Ottawa, CANADA K1Y 4H7
---------------------------------------------------------------------
Michael Covington
: Fernando Pereira <8414-42463> 0112220 (per...@alice.att.com) wrote:
: : What matters in the business world, and even
: Is this "what matters" or "what should matter" in the business world? Sadly,
: it does seem to me, at least here in the UK, that most job adverts ask for
: skills in specific tools, and because of this students are demanding that
: use of those tools is taught in an academic environment even if this is at
: the expense of long-term deeper problem-solving skills. There's pressure to
: take "it's good for you, but you might not appreciate it" things like logic
: programming out of the curriculum for fear that they put off good potential
: students.
It is practically a truism that business doesn't know what is good for itself.
You can hear practically any economist or management scientist say so.
I'm in favor of a relevant curriculum, but not in favor of spending a lot of
time teaching things that will be out of date in a couple of years. And of
course the whole reason for studying theoretical CS, if you are going to be
employed in the field, is to be prepared to understand the computers and
software of 5, 10, or 15 years from now, not just those of today.
--
U54...@uicvm.uic.edu
in the private sector:
Businesses (in America, anyway) are interested in cost-sharing with
other entities, be they governments, other businesses, whatever.
Hiring employees with well-developed and limited skill-sets does not
hurt them in the long run. The skill sets can be taught elsewhere, as
they are for most data-processing skills, and the employees who
cannot adjust to a new skill-set at that firm when skill requirements
change later on are simply discarded. This also costs the business nothing,
as the costs are borne by the individual, his/her family, and society.
Since taxes are inequitably assessed, even this last category (society, or
government, if you prefer) is less costly to business than retraining.
I have seen this process at work at both of my last two assignments,
and was subject to it at my last employer. I am sorry that this is how
things are done. I advocate changing it, but I am afraid that this change
requires a "paradigm shift" in society, approached with great reluctance
by most people (especially those with power and a vested interest).
+--------------------------+--------------------------------------------+
| Nick Geovanis | Consultant |
| U54...@uicvm.uic.edu | Technical Support Group, Inc. |
|--------------------------+ 360 N. Michigan Ave, Suite 1005 |
| What Hath God Wrought? | Chicago, IL. 60601 Tel: 312-704-5100 |
+-----------------------------------------------------------------------+
Fernando Pereira
Fernando Pereira <8414-42463> 0112220 (per...@alice.att.com) wrote:
: What matters in the business world, and even
: in the research world, is problem-solving abilities. At any given time,
: certain tools dominate any given problem class, but what stays much longer
: than the particular tool is the expertise in the problem class, its history
: and evolution.
Is this "what matters" or "what should matter" in the business world? Sadly,
it does seem to me, at least here in the UK, that most job adverts ask for
skills in specific tools, and because of this students are demanding that
use of those tools is taught in an academic environment even if this is at
the expense of long-term deeper problem-solving skills.
It might not be what matters in the short run for a particular job in
a particular environment, but it is certainly what matters in the long
run for the well-being of both employer and employee. The knowledge of
business processes represented in employees' minds and in software is
increasingly the main asset of many organizations. That knowledge
encodes the best current solutions of particular business problems. If
the knowledge is not represented sufficiently explicitly in
conceptualizations, process analysis and design documents and
documented software, the organization doesn't own its future. For
example, a manufacturer that does not own a high-level model of its
inventory management processes, but only an ad-hoc inventory control
system grown over many years, is in serious danger of not being able to
compete with more knowledgeable competitors (such things happen all
the time and have major impacts on profitability). But you cannot
expect good models and designs to be developed by teams that lack
logical analysis skills. Just learning to be a good C++ hacker, say,
does not automatically impart those skills, as I have seen over and over
again in business organizations that I know well.
From the employee's point of view it is also very dangerous to
overspecialize, as I noted in my earlier posting. It is valuable to
know intimately certain widely used tools, but without higher-level
skills one has great difficulty in using those tools effectively in
new problems, or learning new tools, let alone inventing them.
I've been working with computers for 20 years (yikes!). I've known and
used in sufficient detail for systems programming 6 operating systems
(not to count many Un*x variants), 10 or so programming languages, 10s
of text editors and formatters, 100s of utilities, etc. I see that
knowledge as similar to knowing how to get to work from home:
essential, maybe even enjoyable, but shallow and fungible. Once upon a
time I knew how to use all non-privileged TOPS-10 system calls. I
can't even remember now the names of more than a few. In contrast, the
knowledge of mathematics, CS theory, algorithms and programming
structures that I have acquired in the same period may become rusty
when not used, but is always there to serve as a foundation for new
knowledge and to be restored when needed. That is the knowledge I have
rightly trusted to support my working life. If I had entrusted my
career to a few programming systems and tricks rather than to deep and
reusable knowledge, I would have had nowhere to go pretty quickly.
If business really does agree that what matters is problem-solving abilities,
I wish it would do a bit more to make this clear to the next generation so
that they are motivated to take this sort of thing seriously when academics
want to to it to them.
What will get a CS major in the door of a company is specific
knowledge of the most widely tools and practices. What will keep her
there and make her successful is her problem-solving ability,
adaptability and ability to learn. Universities need to structure
their majors to satisfy those two requirements. There is some
conflict, as is always the case when trying to achieve several goals
with bounded resources, and the balance should be different for
different students and career plans. But a university that is
preparing people for a life of work cannot afford to just teach them
the tricks of the day. Achieving those two goals effectively may
require major restructuring of teaching, in particular of the more
theoretical or speculative topics. There may be no room for a "logic
programming" course in an undergraduate curriculum, but there is
certainly room for logic programming ideas as they relate to different
aspects of logical problem analysis, programming, information
representation and algorithms.
--
Leon Sterling
interesting reading, especially the posts by David Cohen, Fernando Pereira
and Matthew Huntbach.
I appreciated Fernando's idealistic comments about the need for general
skills and I have met one or two recruiters that share his views. For
example, someone from Hughes (or some such US company) once told me that
he doesn't hire computer scientists because they are too narrowly based,
but prefers mathemeticians and physicists because they have better problem
solving training and can easily acquire specific computer skills.
Having said that, my impression is that what Matthew says about the UK is
true for the US.
>Is this "what matters" or "what should matter" in the business world? Sadly,
>it does seem to me, at least here in the UK, that most job adverts ask for
>skills in specific tools, and because of this students are demanding that
>use of those tools is taught in an academic environment even if this is at
>the expense of long-term deeper problem-solving skills. There's pressure to
While Prolog is not a turn-on these days, at least one student told me that
having Prolog down on his skill set (he did a Prolog project in an AI
class) was a definite asset in his job interview and led to him being
hired. I'm not surprised that Michael Covington says that his graduates
have no problem getting jobs. Having an extra skill should never be a
disadavantage, but placing yourself as a 'Prolog bigot' is probably not a
good job hunting strategy.
I cannot resist responding to charges of drinking spiked tea as below.
>> No less than Leon Sterling wrote in his introduction to the second edition of
>> "The Art of Prolog" that Prolog seems to be doing so much better since the
>> first edition came out and that this is evidenced by such indicators as the
>> Prolog 1000, a list of 1000 real Prolog applications. I think someone must be
>> spiking his tea. As documented above, Prolog is not thriving and the Prolog
>> 1000 to my knowledge has not yet even reached 1000 and of course many of
>>these
>> applications were done overseas, meaning that they imply nothing about the
>> American marketplace and even if 1000 such applications were done, this is
>> about as meaningful as advertisements from software vendors suggesting that
>> major universities and companies have purchased their products. So what? So
>> what if 1000 applications exist if no one knows or cares about them except
>>the
>> denizens of the Prolog ghetto? And beware seeing
>> the same projects mentioned in articles in magazines written by the same few
>> people. More evidence of the ghetto.
>>
A motivation for starting the Prolog 1000 was sharing the experience I was
having hearing about new Prolog applications wherever I went. And being
tired
of answering the question (usually defensively) 'What has Prolog been used
for?' So if there were a documented collection of successful Prolog
applications everyone would be better off. Almost 700 applications have
been collected so far. The only reason that over 1000 are not on the Prolog
1000 is the lack
of funds for someone to do the job properly of cataloguing the
applications. In contrast to what was written, the applications on the list
are not known only to
people in the ghetto.
Indeed, I continue to hear about new Prolog applications. For example, the
latest Innovative Applications of AI conference had two papers on
applications written in Prolog. (neither of which I'd previously heard of.)
EDS : rule based system to assist re-engineering of a legacy application
for
Federal Aviation.
US Customs: system for assiting inspectors for smuggling protection.
Embodies
rules in Prolog and Oracle.
I am currently editing a special issue of JLP on applications. What strikes
me about the submitted papers (and the papers at the Practical Applications
of Prolog conference) in contrast to 6 years ago when the last special
issue on
applications was put together was how much more mature the applications
were.
There are real applications out there making money. However, it is a hard
fight getting agreement to implement final systems in Prolog - there is a
lot of pressure to reimplement in C or C++. No easy answer to that.
Actually I am not as optimistic on the future of Prolog now as compared to
January 1993 when the preface to the second edition of 'The Art of Prolog'
was written. The most worrying feature is the fate of the Prolog vendors.
Some have disappeared, some are in trouble, some are surviving, none are
thriving. Not having a long-time commercial supplier of software is a
severe impediment to that software being used in industry. Perhaps large
companies will pick up the slack.
The future will be on hybrid systems. The most successful applications I am
reading about are mixtures of C++ and Prolog. What we need to ensure is
that people are educated on the best tools and ideas to use. I have no
doubt that logic programming will remain an important idea.
Leon Sterling Phone: (216)
368-5278
Department of Computer Engineering and Science Fax: (216) 368-2801
Case Western Reserve University email: le...@ces.cwru.edu
Cleveland, OH, 44106
Alan Newman
co...@zurich.ai.mit.edu (David Cohen) writes:
Dear Ahmed,
I hope you will forgive me for taking so much space to answer your
simple question about the availability of Prolog jobs. I hope that this
reply of mine will help guide you and others in your position.
Introduction: Some Facts
------------------------
First off, before I offer you my own, admittedly subjective experiences and
the subjective opinions they engendered, let me present you with some facts:
Examine the want ads in the newspaper and you will find no Prolog jobs.
I see them occasionally. However, that is not what you should be looking
for. If an employer wants a generic programmer, they will be asking for
C. Knowing Prolog is not enough. You have to know how to be productive
for an employer. Prolog is a tool, not an end in itself. Employers look
for solutions to problems. The want ads in the newspapers generally
describe the problem domain, and *minimum tool expertise* the employer
expects from an applicant. If the problem is hard enough, and you have
the skills to solve it, and Prolog is the right tool to support your
efforts, then most competent employers will be interested in you more so
than *not* using Prolog!
Call around to various job recruiters and you will find no Prolog jobs.
In fact, were you to examine all these sources over the last 5 years as I have
done, you will still find Prolog jobs as scarce as hens teeth.
You shouldn't be looking for *Prolog* jobs, unless you want to work for
a Prolog vendor. I like using Prolog, and for the last 9 years, I have
had no problem finding jobs with where using Prolog was an asset. I have
used Prolog to build diagnostic systems for military communication
shelters as well as for manufacturing equipment; automatic test equipment
language translators; graphic systems; and am currently building a legal
reasoning system that is already in use by the company's attorneys. Only
in the case of the graphic system was Prolog a requirement. Each of the
others were problems waiting for solutions, not waiting for a Prolog
programmer. The job classified ads are there mostly because an employer
has a job waiting for a solution, not because an employer likes to *do*
Prolog.
Perhaps you will find a half decent job involving Prolog. Congratulations.
Say it lasts and the company doesn't collapse, like my former employer BIM
Systems US did (only the US company--they are OK in Belgium). But then a little
while down the road you want to move. If you haven't developed some solid C,
C++, Smalltalk, telecommunications, client-server, or other commercially
viable skills, you had better be prepared for a long job hunt since employers
are being very picky and are not likely to give you extra points for knowing
an exotic language like Prolog. If you don't know SQL or possibly more
correctly, if you don't know Sybase or Oracle, the fact that you know all
about relational database and set theory from your Prolog experience will not
amount to much.
Let me rephrase what you just said: If your only skill is Prolog, you
have problems.
The difference between Prolog and C is that large employers staffing up
for large team projects tend to hire most of the fresh graduates where
tool skills are their main strengths. Those jobs mostly require C, and
rarely Prolog. A senior level employee is task oriented, not tool bound.
When you have the ability to design solutions to major problems, you will
often have the opportunity to apply Prolog... if it is the right tool for
that problem.
Reality Check
-------------
Am I perhaps alone in feeling this way? Am I just a disgruntled ex-Prologer
who sold out to the establishment? Could be, you don't know me after all.
Yes, David, I do know you. :-)
You just haven't reached the point where employers depend on *you*
to solve big problems. Unfortunately you have concentrated on the
tool, not broader problem solving skills. When someone else is
designing a solution, they get to pick the tools, not you. I have
a hunch, David, that you'll be using Prolog again.
Regards,
Alan
/\ /\ /\ /\ /\M. Alan Newman, Motorola, Scottsdale AZ /\ /\ /\ /\ /\
\/ \/ \/^^\/ \/ al...@geg.mot.com Speaking for myself.\/ \/ \/^^\/ \/
Fernando Pereira
> The difference between Prolog and C is that large employers staffing up
> for large team projects tend to hire most of the fresh graduates where
> tool skills are their main strengths. Those jobs mostly require C, and
> rarely Prolog. A senior level employee is task oriented, not tool bound.
> When you have the ability to design solutions to major problems, you will
> often have the opportunity to apply Prolog... if it is the right tool for
> that problem.
pretty much the same point, that tool expertise is what is asked of
entry-level candidates, but that conceptual and problem-solving skills
are what is important *in the long run* for a career and for employers
that understand the value of their investment in employees that
understand their business. Even more curiously, that purportedly naive
position was presented independently by the two people with most
industrial experience in this discussion... One wonders to what extent
some academics may be going overboard fine-tuning their course
offerings to the short-term needs of entry-level jobs and thus
weakening the value of the education they provide as a long-term tool
for learning and problem-solving yielding much higher lifetime
benefits both for students and industry.
A few days ago I looked by chance at a recent issue of "Fortune" in
which the cover story was on the ascent of "technicians" in successful
US companies, meaning to them those people that can maintain, debug
and improve on a day-to-day basis the core processes of businesses
(eg. network administrators, data analysts, process engineers). The
message of the article was precisely that such jobs are increasingly
central to successful competitive businesses, and they require a
combination of tool skills, problem-solving skills and knowledge of
business processes that only employees experienced in the business can
have (with the help of suitable education and training).
I believe the picture is very relevant to CS jobs in general. The NAS
report "Computing the Future" (which was much criticised, partly
fairly -- it was a hasty and somewhat disjointed job, and partly
unfairly -- it targeted many well-deserving sacred cows of academic
CS) emphasized the importance of a new style of CS education that
would combine creatively CS technique, tools and understanding of
CS-related problems and problem-solving approaches in various areas of
human activity, ranging from genetics and health-care to traffic
management and flexible manufacturing. The present thread arose from
the clash of two strawpeople, the "academic CS type with exotic skills
such as logic programming" and the "hard-nosed business person who
just wants the largest number of lines of C++ written for the lowest
cost". This clash of midgets hardly reflects what is really going in
academia or industry, let alone the value for careers and businesses
of combining a rigorous CS education with deep understanding of
business problems and processes.
Sandiway Fong
|> Where Prolog Seems to Be Heading--My Experience
|> -----------------------------------------------
|> The response that I got back was primarily that either the Prolog
|> was gathering dust or that whatever system they had written in
|> Prolog was now ported to C.
I wonder how much of this is due to the fault of Quintus and perhaps
some other vendors? One of the big differences between C and Quintus
Prolog programs is that you can distribute *customizable* programs and
object code for the former, but not the latter. (Yes, I know about
run-time systems but they don't allow recompilation.) On the C side, a
compiler either comes bundled or you have a freely available very full
featured C (gcc) at your disposal. However, the free Prologs aren't
anywhere as good or full featured as Quintus. Quintus then cost
$10,000 per (node locked) license for non-academic sites (I know now
it's $5,000) which practically guaranteed very low availability. So,
$10,000 vs. almost zero; no contest. (I know you can also take the
route of paying a few K for a Sun C compiler under Solaris 2.x.)
Now, Quintus is about to loosen its conditions slightly, and the
pricing barrier has been halved. This is good, and Quintus is to be
applauded, but IMHO perhaps this should've been done a long, long time
ago. Today, I'm afraid. even $5,000 per developer is too much for
most institutions and I wouldn't be surprised if Quintus doesn't see a
big boost in sales.
Sandiway
---
Dr. Sandiway Fong
NEC Research Institute Inc.
4 Independence Way
Princeton NJ 08540
Sandiway Fong
|> Reality Check
|> -------------
[text deleted]
|> This leads me to my conclusion which is, while things may be somewhat
|> different
|> in some other countries such as England, where Prolog is bigger, things here
|> are very grim if one is looking for a Prolog job. I must say that I am very
|> displeased to see members of the academic community try to suggest that things
|> are otherwise, since encouraging people to seek their fortunes in the Prolog
|> marketplace as it now exists is just short of criminal.
I agree with David Cohen here. The pyramid is just too small, if the
numbers of users of Prolog programs aren't worth a blip on the global
radar screen, then the number of jobs for development (or research)
work is obviously only going to be a small fraction of that. But
actually, I would guess that there are fewer users than developers, a
sign (if we ever needed one) that Prolog hasn't taken off. I'm
thinking that it's probably the case that most Prolog program users
are actually in the class of developers.
I also agree with the other posters who say that Prolog is just
another skill (a bonus if you like) to be picked up along with the
rest (algorithms, theory, C etc.) Although I am in need of people who
have great Prolog programming and debugging skills (which can only
come from lots of experience), I would never hire someone who couldn't
also do C etc. The world extends beyond the Prolog prompt, at least on
Unix machines...
Sandiway
--
Fernando Pereira
> co...@zurich.ai.mit.edu (David Cohen) writes:
> |> The response that I got back was primarily that either the Prolog
> |> was gathering dust or that whatever system they had written in
> |> Prolog was now ported to C.
> I wonder how much of this is due to the fault of Quintus and perhaps
> some other vendors? One of the big differences between C and Quintus
> Prolog programs is that you can distribute *customizable* programs and
> object code for the former, but not the latter.
> Now, Quintus is about to loosen its conditions slightly, and the
> pricing barrier has been halved. This is good, and Quintus is to be
> applauded, but IMHO perhaps this should've been done a long, long time
> ago. Today, I'm afraid. even $5,000 per developer is too much for
> most institutions and I wouldn't be surprised if Quintus doesn't see a
> big boost in sales.
grammar-compilation tool for speech recognition I wrote partly in
Prolog, and that has not been nearly as used as it could have been if
I could give the code away to interested parties within AT&T without
requiring them to buy a Prolog license. Also, when I was involved with
Quintus this was a sticking point with many customers, and I argued
constantly against what I thought were self-defeating pricing
policies. Sure, high prices and lack of portability created an
early burst of income that was misinterpreted as a steady stream, but
soon users with real applications started getting tired of all the
obstacles, including high delivery costs, to diffusion of their
applications. Yet another instance of the well-known phenomenon in
which sort-term optimization of income may be bad in the long-term.
More generally, there's a question as to how much premature
commercialization may get in the way of the spread of new programming
languages. In the functional programming world, for instance, it is
instructive to compare the relative penetrations of Miranda and
Standard ML. Miranda became early a commercial exclusive, while SML
has at least one high-quality free compiler, SML/NJ. Many universities
and some companies are now using SML, but very few use Miranda (Yes, I
do understand that this may be also due to the differences in
efficiency etc. between a purely functional language like Miranda and
SML with its imperative features, but I think free availability is
more important at least with respect to use in education). Good
Schemes are also free. Once a language has a substantial loyal
following who has come to depend on it in their work, there is a
market for commercial versions with better features. However, before
that point, commercialization may suck in the few capable
implementers, and a product is created for which a market doesn't
quite exist yet.
Merritt
I learned Prolog as a hobby, but then realized its potential where I
worked, so I went to my management and told them the neat things I could
do, which solved real business problems, and they said go do it so I
suddenly found myself working fulltime with Prolog at work.
In particular, I was working at Cullinet software, which had a major
problem with an overstretched field support organization. I proposed two
expert systems, one which did tuning of their IDMS product, and the other
which did installation.
As is often the case, the biggest problem was getting the time of the top
field support guy, D.C., to build the tuning system. After being blown
off by his management a number of times I expressed my frustration, and
D.C.'s manager yelled at me "If I could clone D.C. I'd love to, but I've
only got one D.C.", to which I replied, "that's exactly what I'm trying
to do." I got the support and in a few weeks we had a system that tuned
IDMS parameters better than 90% of the field support staff.
Footnote: but Cullinet was failing rapidly at the time and went under before
the systems were ever formally deployed. Maybe it was Prolog? (I hope
not, I'm still trying to make a living with Prolog as a vendor now,
trying to come up with ways for ordinary programmers to sneak a little
Prolog into their applications.)
Dennis
Merritt
Having been a Prolog user, and having subsequently acquired a Prolog, it
is our intention to provide a solid commercial-grade Prolog that is
affordable on individual as well as corporate budgets, that has no
runtime fees, that is portable across multiple platforms, and that is
easily accessible to the C/C++ programmer.
That is, I want to provide a Prolog that will make the kind of transition
I had into Prolog (from a commercial programming environment) easier for
others.
Dennis
Michael Covington
: From article <leon-200...@dvorak.ces.cwru.edu>, by le...@ces.cwru.edu (Leon Sterling):
: > There are real applications out there making money. However, it is a hard
: > fight getting agreement to implement final systems in Prolog - there is a
: > lot of pressure to reimplement in C or C++. No easy answer to that.
: Surely, one easy answer is that reimplementation in C/C++ raises
: maintenance costs [...]
: I can understand wanting to reimplement a Prolog program in machine code,
: but then we have compilers to do that cheaply and reliably ...
I think you have hit the nail on the head. There are a lot of fuzzy-
thinking people out there who think C *is* machine code. :)
Peter Ludemann
>for an employer. Prolog is a tool, not an end in itself. Employers look
>for solutions to problems. The want ads in the newspapers generally
>describe the problem domain, and *minimum tool expertise* the employer
Would that this were true. There are some employers who are
interested in problem solvers. But the attitude nowadays seems to be
(exaggerating only a little) "if you can use a hammer, you must be a
carpenter; and you can no doubt also design houses".
Even in fairly small companies, you need to get past the "human
resource" people, who often have little understanding of the job needs
and usually look for certain key words in people's resumes. Exotica
such as "Prolog" mean nothing to them, except perhaps time wasted when
one coul have instead developed expertise in C++.
When I went through university (he says, stroking his grey hairs),
nobody knew what the ideal programming language or operating system
was, so we had to learn two or three dozen programming languages and a
dozen operating systems. Today, the good news is that we know what
the ideal programming language and operating system are; the bad news
is that they're C++ and Unix (or Windows-NT for Microsoft junkies) [I
think that Gabriel said this first].
On the other hand, Ullman's book on databases contains a lot of
Prolog-like stuff to describe relational databases; and I see that the
C++ people are slowly realizing that garbage collectors are awfully
handy for even conventional programming (old-timers will remember that
Algol-W and XPL were fairly conventional 3GLs with garbage collectors;
they didn't seem to suffer noticeably from inefficiency but I suppose
each generation needs to rediscover certain things for itself). So,
perhaps Prolog as such will fail; but it could live on in "deductive
databases" and some C++ class libraries, although probably with a much
uglier syntax than the Prolog that we know and love/hate today.
I don't regret the time I've spent learning and using Prolog, even
though I know that the people who take over my project will
immediately rewrite it in yacc/lexx/C++, at 10 times the cost of my
original work. After all, I had the fun part, where I could
concentrate on the design issues without worrying about my tools
getting in my way.
--
Peter Ludemann Senior Computer Scientist
Institute for Medical Informatics InterNet: lude...@imi.kdsi.com
124 University Ave., Suite 350 Phone: +1-415-325-5178
Palo Alto, California 94301 FAX: +1-415-325-5177
Paul Singleton
> There are real applications out there making money. However, it is a hard
> fight getting agreement to implement final systems in Prolog - there is a
> lot of pressure to reimplement in C or C++. No easy answer to that.
Surely, one easy answer is that reimplementation in C/C++ raises
maintenance costs:
when (as is inevitable) the design is revised,
either you modify and retest the Prolog version, then expensively
redo the reimplementation in C/C+,
or you riskily and short-sightedly hack the derived C/C++ code,
leaving you without a Prolog version.
If your application requires backtracking, unification, dynamic memory
allocation and garbage collection, then it is surely bad software
engineering to prefer to manage a program in which the implementation
of these is intimately intermingled with the application-specific code
rather than use a system in which they have been neatly abstracted out
into a well-verified (if not formally proven :-) kernel.
I can understand wanting to reimplement a Prolog program in machine code,
but then we have compilers to do that cheaply and reliably ...
----
__ __ Paul Singleton (Dr) email: pa...@cs.keele.ac.uk
|__) (__ Computer Science Dept. tel: +44 (0)782 583477
| . __). Keele University, Newcastle, fax: +44 (0)782 713082
Staffs ST5 5BG, ENGLAND road: M6 J15 follow signs
Peter Ludemann
> Where Prolog Seems to Be Heading--My Experience
> -----------------------------------------------
[text deleted]
> for a long time, but now they have even stopped advertising.
Check out Quintus's new ad (in AI Expert). Nicely laid out (speaking
as somebody who once worked with typesetters). The headline is:
"Quintus Corporation Celebrates Its TEN YEAR Anniversary". And it
emphasizes "... a suite of powerful tools that work together,
integrate into C and C++ worlds ...". A definite improvement over the
ads I used to see. When will they target the moaning masses in Byte,
Dr. Dobbs, and PC-Magazine, I wonder? Perhaps soon: they're even
mentioning "Special Anniversary Pricing".
In a world where perl, tcl, GUI-builders, SQL and various other tools
happily co-exist with the C/C++ hackers, if the various Prolog vendors
can come up with a good marketing strategy, Prolog could be around for
a long time. [On odd days I think that C++ is the end of programming
language evolution; but on even days I think that people will
eventually desire something better.]
> I called [Quintus] recently to check. Everyone I named including
> long-time star employees such as Tim Lindholm were gone.
There are still quite a few star employees around (you probably just
don't know them); and those people would be super-stars at many other
companies IMHO. I'm willing to bet that the people staffing other
Prolog companies are also well about the average "programmer".
Quintus has been trying to develop other markets by leveraging Prolog
technology into "conventional" places where it hadn't been previously
tried. It now looks as if they're using some of that experience to
re-think their strategy for selling Prolog. I noticed that ALS is
selling genetic and neural networks. Will the other vendors come up
with new marketing strategies, too? We can only hope ...
Peter Schachte
> reimplementation [of a Prolog prog] in C/C++ raises
> maintenance costs: when the design is revised, either you modify the
> Prolog version, then redo the reimplementation, or riskily and
> short-sightedly hack the derived C/C++ code.
I agree completely. Unfortunately, this argument presupposes a
different mental model of what is going on than a manager is likely to
have. The typical software development manager is likely to view the
Prolog program as a prototype, and the C++ reimplementation as the
"real" program. The sooner the embarrassing (because the C++
reimplementation took so long) and uncomfortable (because the manager
doesn't understand it, nor do many of the engineers) Prolog version is
forgotten, the better.
>If your application requires backtracking, unification, dynamic memory
>allocation and garbage collection, then it is surely bad software
>engineering to prefer to manage a program in which the implementation
>of these is intimately intermingled with the application-specific code
>rather than use a system in which they have been neatly abstracted out
>into a well-verified (if not formally proven :-) kernel.
True enough; a canned solution is more attractive than rolling one's
own. Still, it may not be attractive enough to overcome inertial
resistance. All of these problems can be solved, with some pain, with
the familiar tools. As the saying goes: if all you have is a hammer,
everything looks like a nail. A corollary of this is that if the only
tool you know how to use is a hammer, everything still looks like a
nail, even if you have a full toolbox at hand.
If Prolog is ever to become a popular language for commercial
application development, it will happen very slowly, one success at a
time. Unless, of course, it is made backward compatible with C++ and
renamed C*=C :-).
--
-Peter Schachte
pe...@cs.mu.OZ.AU
Ian Dickinson
I have been following this thread on and off for a while now, and some
interesting points have been debated. Rather than continue the debate, I'm
going to offer a much more mundane contribution: jobs which mention prolog
as a desirable or required skill are often listed (besides a lot of other
useful stuff) on Ken Laws' "The Computists Communique" list. It costs some
$$ to subscribe, but over the past two years or so I have found it well
worth the money. There is a special student rate. A quick count in the
last issue reveals two jobs mentioning prolog as a skill out of eleven
AI-ish opportunities listed (Europe & USA). Further details from:
>Publisher/Editor: Dr. Kenneth I. Laws, 4064 Sutherland Drive,
> Palo Alto, CA 94303, USA. Phone: (415) 493-7390.
>Internet: la...@ai.sri.com (courtesy of SRI International).
I have no connection with TCC other than being a very satisfied customer.
Enjoy the journey!
Ian.
--
| Ian Dickinson Hewlett-Packard Little Falls Site Wilmington, DE, USA |
| Email: i...@lf.hp.com | Enough if something from our hands have power|
| These opinions are mine, | To live and act and serve the future hour |
| and not (necessarily) HP's | - Wordsworth |
Alan Newman
pe...@munta.cs.mu.OZ.AU (Peter Schachte) wrote:
If Prolog is ever to become a popular language for commercial
application development, it will happen very slowly, one success at a
time. Unless, of course, it is made backward compatible with C++ and
renamed C*=C :-).
Renamed C*=C? Maybe :-C would be more apropos. :-)