|
|
|
|
|
|
|
|
Are there Prolog Jobs out there?Keywords: jobs.
From: bouzid@csgrad.cs.vt.edu (Ahmed Bouzid) 14th August 1994I have been programming in Prolog for the past six years -- all kinds of flavors, versions, etc. -- but all in a 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?
From: mkant+@cs.cmu.edu (Mark Kantrowitz) 15th August 1994I've added the following to the Prolog Resource Guide. There are similar mailing lists for lisp-jobs and ai-jobs.
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+prolog-jobs@cs.cmu.edu Resumes should not be sent to the list.
To subscribe, send a message to ai+query@cs.cmu.edu with:
subscribe prolog-jobs FirstName LastName, Affiliation/Organization
in the message body.
For help on using the query server, send mail to ai+query@cs.cmu.edu with:
help
in the message body.
If you have any other questions, please send them to ai+@cs.cmu.edu
From: cohen@zurich.ai.mit.edu (David Cohen) 17th August 1994I 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 and you will find no Prolog jobs.
Examine the postings in misc.jobs.contract and 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 were not able to point me to anything and they themselves don't often have openings, since they are 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 the comp.lang.prolog newsgroup. Even Quintus, the flagship Prolog company, has not been pushing Prolog for a long time, but now they have even stopped advertising. 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.
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 dc@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.
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.
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.
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.
And so to my conclusion: 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 trying to suggest that things are otherwise. 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!!!!"
From: pereira@alice.att.com (Fernando Pereira) 18th August 1994David Cohen 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. 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 taught. 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.
From: mcovingt@aisun5.ai.uga.edu (Michael Covington) 17th August 1994David Cohen wrote:
"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.
From: cohen@zurich.ai.mit.edu (David Cohen) 18th August 1994No, no, no and no. A thousand times no. The analogy might be appropriate for a COBOL programmer or some such, 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.
The companies not doing Prolog do not care. They do 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.
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.
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.
From: mmh@coffee.dcs.qmw.ac.uk (Matthew Huntbach) 19th August 1994Fernando Pereira 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. 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 teach it to them.
From: leon@ces.cwru.edu (Leon Sterling) 20th August 1994I 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.
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.
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.)
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, is how much more mature the applications are. 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.
From: alan@geg.mot.com (Alan Newman) 18th August 1994David Cohen writes:
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.
He continues:
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.
Let me rephrase what you 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.
From: imi-kdsi@nic.cerf.net (Peter Ludemann) 26th August 1994Alan Newman writes:
... 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. ...
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 could have instead developed expertise in C++.
From: daver@quintus.com (Dave Rahn) 23rd August 1994As the Quintus Vice President in charge of Prolog, I would like to clarify what Quintus is up to.
1) Quintus is still a flagship Prolog company. We continue to advertise. Keep your eyes open for the September 1994 issues of AI Expert and PC AI.
2) Quintus Prolog is a profitable business unit. This is true even when the application business is not considered.
3) Quintus is very active in product development. In response to a solid base of customers/prospects who want to do as much as they can in Prolog, we will shortly be announcing a product suite called "OpenACE" (ACE = application construction environment) which includes packages layered around Quintus Prolog. These packages provide support for constructing:
From: sandiway@research.nj.nec.com (Sandiway Fong) 21st August 1994David 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. 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.
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.
From: imi-kdsi@nic.cerf.net (Peter Ludemann) 27th August 1994David Cohen writes:
Even Quintus, the flagship Prolog company, has not been pushing Prolog for a long time, but now they have even stopped advertising.
Check out Quintus's new ad (in AI Expert).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.
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 ...
From: ian@lf.hp.com (Ian Dickinson) 6th September 1994Jobs which mention Prolog as a desirable or required skill are often listed (together with a lot of other useful stuff) on Ken Laws' "The Computists Communique" list. There is a subscription fee, 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 and USA). Further details from:
Publisher/Editor: Dr. Kenneth I. Laws 4064 Sutherland Drive Palo Alto, CA 94303, USA Tel: + 1 415 493 7390 Email: laws@ai.sri.com
I have no connection with TCC, other than being a very satisfied customer.
From: Ken Laws LAWS@ai.sri.com 18th September 1994I had very little success in finding Prolog jobs until this past month, when Mark Kantrowitz started his prolog-jobs list. (Also lisp-jobs and ai-jobs.) Mark has kindly allowed me to summarize his ai-related postings in my Computists' Communique, but serious Prolog job-seekers should subscribe to his list.
The Computists' Communique is an AI/IS/CS news service of Computists International. Weekly issues of about 32KB (8 pages) include job ads, journal calls, NSF announcements, grant and research news, online resources, and business tips. You get a concise, time-saving summary from dozens of sources.
Two less-condensed digests are also available. One covers job opportunites that are too applied to be summarized in the weekly Communique; the other forwards research software announcements. These "press release" collections are free to anyone who wants them.
Sample issues of the Communique are available -- please ask for one if you're at all interested in joining.
First-year dues are $135 for professionals, $55 for students (or those with equivalent salaries), and free for unemployed computer scientists. Members outside the US get an additional 50% discount. (Renewals may also qualify for a 20% discount.)
|
|
|
|
|
|
|
|