Are Perl and the dynamic languages dead or what ?

[szabgab]

Reading this report about tech job rise, based on the June report of Dice.com you might learn that the demand for Perl skills dropped by 12% since the beginning of the year.

On the other hand jobs.perl.org seem to indicate a 50% growth from the beginning of this year (see stats). Articles such as this and this and the Dynamic Languages Symposium seem to think that the time of dynamic languages has just came.

Most of the people I talk to also seem to think that Perl is dead and they hardly ever heared about the other dynamic languages. (This might indicated I am not talking to the right people though).

I think I can assume that the majority of the Monks also think that there is a bright future in dynamic languages, especially in Perl.

So what do you think why is this gap between what Dice sees and what the those other articles try to tell or what we would like to belive ?

[brian_d_foy]

Any such change is more likely to less people are using Dice and more people using jobs.perl.org. To post a job to Dice for 30 days, you pay over $350. To post to jobs.perl.org, you fill out an email. It's hardly surprising that the one that costs money is losing.

The trouble with all these reports is that they don't explain, or can't explain, what a job is. Did they remove the same job posted multiple times because it could not be filled or because they posted it with different text to target it to different people? Was Perl merely listed as a skill among many in the shotgun approach, or was it the major skill the employer needed? Was the job actually a Perl job, or was it a bait-and-switch (which I've seen enough to not discount)? What is 12%, anyway? Let's see the hard numbers and the long term fluctuations.

However, a drop in job postings is not a drop in demand for Perl skills. It's a drop in demand for new employees with that skill. They still keep their current employees who still have that skill. I get to talk to a lot of very big companies as I try to sell Stonehenge services, and the most common answer I get is that they aren't hiring new people (so they don't need Learning Perl courses). To put the numbers into perspective, I want to put them next to the general trend in hiring.

Perl is far from dead. Instead of job postings, which I consider more an indicator of economic cycles, I like to look at the activity on CPAN. There are several new modules every week, and there are several new people asking for PAUSE IDs every week. People are doing things in Perl.

Business are still putting their money behind Perl too. At Stonehenge, we've had to expand quite a bit just to handle the load this year. O'Reilly is putting money into a new edition of Learning Perl which comes out next month. Apple Computer invited Randal and I (among others) to speak at WWDC about Perl. TPF got a $70,000 grant from NLnet for work on Parrot.

Perl doesn't seem dead to me.

[jhourcle]

The Perl jobs from the sorts of companies that would advertise on Dice are down. The Perl jobs from the sorts of companies that would advertise on perl.org are up.

What does this mean? I can only assume -- the types of folks that advertise on dice/monster/hotjobs/etc tend to be from outside the 'perl community'. They might be tech companies, or recruiters, etc, but they have no clue where to search for employees with a particular skillset, or they might want people with more broad skills. (I'm of the opinion that overspecialization is bad -- learn a few more languages, so that you have more than just one tool that you can fall back on -- they all have different niches that they fill).

The types of groups that post stuff on perl.org probably don't want folks who have written two scripts, and so consider themselves to be 'perl programmers' -- they tend to be looking for someone with more advanced or more specialized skills.

To some degree, it's a matter of knowing where the fish are -- if you're looking for carp, you fish where there's a lot of carp, not in some place where there's 1% carp, and you're going to have to throw back the majority of your catch.

But, then again, we have no idea how many of the jobs posted on dice are also posted on perl.org, so it may just be that people are getting fed up with the results from dice, and so aren't posting there anymore, and are moving to perl.org, or that they're cross posting, and there really is fewer job openings.

But fewer job postings doesn't mean that there are fewer jobs. It may be that perl programmers have less turnover -- it may be due to job satisfaction, or just a fear of finding a new job in a depressed market. It may be that social networking has made it so that jobs openings are being offered through other channels.

I'm not much of a believer in statistics, unless I know what all of the assumptions are being made, the possibilities of any bias in the results, and how all of the numbers are calculated.

What was the 12% decrease? The overall percentage of programming jobs that mentioned Perl? (ie, it went from 8% to 7%), or in the overall number of job offerings mentioning Perl? (from 800 to 700)? In the first case, if the total jobs increased 26%, and the relative Perl mentions went down 12%, that's still an increase of almost 11%). It's possible to put almost any spin you want on the numbers to prove a pre-determined conclusion.

If you like doing it, and you like your job, then keep doing it. I'd still make sure you're well rounded should things ever go wrong, but I wouldn't overly stress about these things.

[tlm]

There are three kinds of lies: lies, damned lies, and statistics.

Benjamin Disraeli

Trends in the demand for Perl programmers is, of course, a topic of great interest to us in the Monastery, but the sources you cited don't give me enough to form a solid opinion one way or the other. It is certainly possible that demand for Perl programmers has slumped since the beginning of 2005, but the data offered by Dice is too sketchy to be taken seriously. And the support you cited for the "opposite opinion" seems to me even weaker.

[davidrw]

Don't forget this famous quote by Homer:
        people can come up with statistics to prove anything ... Forty percent of all people know that.
But yeah, you and jhourcle and brian_d_foy make good points about the stats.. I'm sure there's other people here that always view any kind of statistical statement with immediate skepticism (sample size? source? misleading/vauge? etc)

[neniro]

Managers never heard anything about perl and such stuff. They hire Java-programmers, cause java is industrial-standard and they know onething: They won't get fired for employing java-programmers (even if their projects go straight to hell).

Let's say that there are a lot of jobs in small to medium sized companies waiting for perl-programmers but didn't know it, they want just someone who gets the job done. If you'll do it in perl they don't care, even if they have employed you as a java-programmer.

[brian_d_foy]

In fairness to managers, once someone makes the decision to use some particular language, that's the sort of programmer they hire. It's not that they don't get fired for hiring Java programmers, but that somewhere else someone decided to use Java. That's not an a priori bad decision, so I don't sweat that.

The other interesting note is that .NET and HTML were the big winners in the Dice report (and PHP was curiously absent, and I would really like to see those numbers). Since .NET was up 52% (but hey, going from 1 to 2 will do that), I'm curious about why that is.

What I'd really like to see from the tech jobs is a breakdown of what the tech workers are actually doing instead of what they are using. Since Microsoft pretty much rules the desktop for corporate workers, I expect that workers supporting those workers and their machines would be much more focused on .NET, Java, Visual Basic, and so on. I don't expect Perl to even be close to dominating that arena (just like I don't expect Linux to).

Once we segment "the market", I don't think these numbers will be all that surprising, or worrisome. Either way, I don't blame the managers, and I think using them as a scapegoat doesn't help anyone.

[neniro]

Of course you're right that it isn't right to blame all of them as I did. But it's my personal opinion that a lot of people in management positions try to go the "safe" way. Nothing wrong about this, but in fact they don't decide they just follow the masses. Thats a general problem not just in programming. They are paid to find good solutions, but they just try to reduce the risk of their decisions and take the average solutions. That can be extremly annoying.

[poqui]

"In fairness to managers, once someone makes the decision to use some particular language, that's the sort of programmer they hire."

Not necessarily. I have worked at companies where even though the majority of the work is in Java, the management still hires ASP programmers or vice versa.

We have requirements for Data Warehouse professionals, and we interview a lot of people who were managers at their last job and haven't touched code in years.

[cbrandtbuffalo]

Since statistics are generally hard to believe anyway, I'll muddy the waters with some anecdotal info. Based on some postings here on Perlmonks such as Developer::Perl::Find and contact with manager types I know in Perl shops, there are jobs out there. These folks are actually having a hard time finding Perl talent with enough experience. I've heard of a similar situation in Seattle, Chicago, Boston, New York City area, and New Jersey. Andy Lester did a whole session on how to get a job at YAPC and OSCON last year partially to increase the quality of applicants for his openings.

I work in a Perl shop in Buffalo and we recently had some openings. Most of the people who responded didn't have close to enough Perl experience, and that was mainly what we were looking for.

Final anecdote is a Buffalo PM member who moved out to Arizona and within a few weeks had a job in a Perl shop doing all Perl.

Statistically, none of this means anything, but for me personally it's evidence that Perl is quite healthy.

[adrianh]

These folks are actually having a hard time finding Perl talent with enough experience. ... Statistically, none of this means anything, but for me personally it's evidence that Perl is quite healthy.

Depends on the reason. I know some people who're having a heck of a job finding decent COBOL programmers. Doesn't mean that COBOL is a healthy language.

Being able to hire decent developers is one of the major factors in picking a language for a project. I know of one successful Perl based e-commerce server that moved over to C++ because the development team had a roomful of good C++ developers and couldn't find any good Perl developers.

[shiza]

My experience with the market has been similar. It seems that the demand for experienced Perl programmers is greater than the number of experienced Perl programmers who are looking for a new position.

[monarch]

What a CEO wants, and what "those in the know" want, are often poles apart.

Talk to any telecommunications company today and they will swear black and blue that VoIP is the only way forward. "Convergence, convergence, convergence," they will shout. And yes, it's a great idea.. I guess.

Talk to anyone that's worked with telephony for any length of time, however, and all of a sudden squashing a real-time protocol into a data network becomes problematic. And what CEO wants to hear about such things as QoS, jitter, statistical multiplexing? CEOs rate themselves if they understand that VoIP will introduce a bit of delay, thinking that their last phone call to Germany wasn't so bad over the PSTN.

What does that have to do with programming? Well, Perl programmers (in my opinion, and my opinion only) tend to understand pragmatism, which involves looking at the world realistically.. and Perl programmers will often be the first to admit when their language doesn't solve a particular problem (although Perl does solve a lot of problems).

Hence it all comes down to who gets into management. Who wins the next board meeting political show down. Trying to predict this is about as accurate as picking whether Nortel shares will go up or down next month.

We're living in an ever changing world, and two things don't change: politics and ever changing feedback loops.. the world is not a time-invariant linear system.. *sigh*

[dragonchild]

Personally, I listen to people who have made money in the field I'm in, which is programming and specifically web applications. To that end, I tend to listen to people like Paul Graham and Joel Spolsky. Both of them feel that you have to justify using a non-dynamic language because it costs more to do so.

I've gotten jobs through Dice. Every single one was a contracting firm that would literally cut'n'paste the customer's req into the Dice form. Often, I'd see 5 companies advertising the same position¹. Do those count as 5 jobs or one?

Another way to look at it is the number of Fortune 500 companies who would go out of business in a month if their Perl programs stopped working, then compare it to the number of Fortune 500 companies who would go out of business in a month if their XYZ programs stopped working. Seeing as I know of 4 Fortune 500 companies where I've personally worked that would keel over if all their Perl or COBOL programs stopped working, but nothing else ... you do the math.

1. This is actually a problem because you can get into image problems if you accidentally get submitted to the same job by more than one company, especially if you sent your resume in at 3am to 200 postings like I used to do.

[ww]

"Consider the source!"

Much of what's popularly reported as "statistical evidence of (fill_in_the_blank)" comes from parties who are NOT disinterested in the results.

With no intention to impugn either of your sources (given my utter lack of familiarity with Dice.com and lack of evidence about the effectiveness, profitability, etc. of job listing sites such that you cited), I would point out that one could infer that a site listing perl jobs might have an interest in showing a growing market. And one need not infer any deliberate skewing (<begin tongue in cheek> while that may sometimes happen in some extraordinary cases <remove tongue from cheek>): sometimes the views of those specifying the data to collect shape the results.

[jhourcle]

I did consider the source -- and Dice is a company that is trying to make money. They make money by getting people to pay them for job postings. People pay them for job postings, because people know they exist. (and forget that they had that little backrupcy issue). People know they exist, because they keep getting quoted in the press for releasing these sorts of statistics.

Although with the internet, just about any stupid thing will get some mention, controversial things tend to get moved to a location where they can get more attention.

I personally don't trust most of the stuff that I read in the news, if it's been summarized to the point where someone has made a decision for me, rather than giving me the information so that I can make the decision for myself.

In this particular case, they've selectively pulled out a few numbers, and the only information that we have is that it came from something about 'skills needed':

*Please note that data reflects public information posted by dice.com users and customers. For example, 10,545 jobs posted on dice.com as of June 1, 2005 describe Unix as a skill needed for the position advertised. A single job posting may reflect more than one skill, location and type of position (permanent vs. contract); therefore total figures for these attributes may be greater than total jobs posted.

However, as anything as a percentage is a ratio, without knowing what the ratio is based on, it's a useless value.

[YuckFoo]

TIOBE Programming Community Index seems to be as use(ful|less) a measure as any.

YuckFoo

[Anonymous Monk]

I suspect a lot of people are looking to Perl 6 for answers to where Perl is headed. So to some extent, the future of Perl depends on what comes out of Perl 6, and when.

[BUU]

I suspect a lot of people are looking to Perl 6 for answers to where Perl is headed. So to some extent, the future of Perl depends on what comes out of Perl 6, and when.

Ya know, I've heard a couple of people say thing similar to this, but I've yet to actually hear anyone say "I'm not going to learn/use/whatever perl5 because perl6 might come out someday". Most people, especially if they don't know perl5, haven't even heard of perl6. So maybe in 10 years people will be phasing out their perl5 scripts and looking for perl6 programmers, but even if perl6 never comes out, I hardly see the demand for perl5 dying out. It's still a great, useful language that's used all over the place, mostly where you'd never expect it. Or perhaps I'm just not talking to the right people.

[monarch]

After all, we're all using IPv4 right.. and IPv6 is supposed to be the next big thing.

The trouble with critical mass is when it's big enough switching over becomes very, very difficult, especially when society has spent years patching something to fix known issues (such as a Boeing 747 with wiring issues, or NAT for IPv4, etc etc).