BitWorking

Tim Bray on Ruby on Rails

Tim Bray fearlessly predicts the future:

Rails will continue to grow at a dizzying speed, and Ruby will in consequence inevitably become one of the top two or three strategic choices for software developers. But at the same time, other frameworks and tool-sets are learning its lessons, so Rails will get some serious competition.

Now you may remember that I made my own prediction about Rails:

...the important point is that Rails peaked somewhere in the middle of 2006 and is now declining.

While I disagree with Tim on the future of Rails, I do agree with him on other frameworks learning the lessons of Rails:

All of that should not diminish Rails, but instead should put it in proper perspective. The real value in Rails was a singular, opinionated vision of how web development should be done. That vision moved the bar for the whole industry, which is quite an accomplishment. But there's nothing magical about Ruby, or Rails. All of the underlying concepts of Rails can be 'ported' to almost any other language, which is exactly what were seeing today, with Rails-like frameworks appearing in Python, Perl, etc. Rails success was defining the platonic solid for second generation web frameworks. Framework 2.0.

The differences between our predictions are that I said Rails would decline and not continue to grow at a dizzying speed - I made my prediction 14 months ago, and finally - my prediction is right. Here is a Google Trends chart for "ruby on rails", in which you clearly see that Rails peaked in the middle of last year.

rails_trend

Tim seems to be suffering from confirmation bias, what with all the Ruby and Rails conferences he's attending, and the company he's working for which is working very hard to find a future life for the JVM. I personally have my doubts about how fit the JVM is for anything but Java. For example, here is a quote off the Factor front page:

Factor began as a scripting language in a Java game project and quickly grew into a general-purpose language. While this was happening, the limitations of the Java virtual machine were making themselves apparent, and an effort to write a native Factor implementation with a minimal core in C was kicked off. The native implementation was bootstrapped from Java Factor, and soon thereafter native Factor became the de facto implementation, work on the Java implementation stopped.

A good check against confirmation bias is to pick an unbiased measure for your prediction ahead of time. I didn't choose Alexa ahead of time for my measure since its numbers can be questionable and open to gaming, but it appears in this case to confirm the Google Trends data.

ruby_on_rails_alexa_graph

So what have we learned? Tim's prediction is already wrong, unless there's some new and unusual definition of "grow at a dizzying speed" that means "shrink". We've learned to take any prognosticating that comes without a pre-selected unbiased metric of success with a huge grain of salt. Finally, we've learned that even the O'Reilly hype machine has its limits and can no longer keep the Rails ball in the air. Maybe next year they'll rewrite "RESTful Web Services" with all the examples in Scala.

Update: Michael Foord in the comments points out a similar flat-lining of Ruby in the TIOBE index. That would be three independent sets of data (Google Trends, Alexa, and TIOBE) that show at the very least that neither Ruby nor Rails are growing at a "dizzying speed". Note carefully that we aren't talking about predictions anymore. We're talking about Tim's statement that Ruby and Rails grew at a dizzying speed this year; the thing you have to believe is true before you can even get to the prediction.

The kinder, gentler Zed.

Posted by Patrick Mueller on 2008-01-04

Touche, Patrick, touche.

Posted by Teltariat on 2008-01-04

O'Reilly hype machine? The most recent Radar post tagged with "Rails" is from September 2006.

Posted by chromatic on 2008-01-04

chromatic,

This search shows over 164 articles on the Radar site that mention Rails. So yes, I'd say hype.

Posted by Joe on 2008-01-04

I think Rails will continue to grow in absolute numbers. For every person that dislikes it, you may find one that likes it. This means that even if Rails grows old and boring, it should keep a certain balance. Also, being able to run Rails on JRuby and eventually on IronRuby should give it another boost or something. If eventually Apache gets it running in an integrated way with a "module", it could grow further. And so on. Deployment problems once fixed could eventually give it another boost. Not that I personally like Rails though, but I don't find the competition appealing either.

Posted by Joao Pedrosa on 2008-01-04

By that logic, the Internet peaked in 2004, and usage of Java is down 50%, and the same for HTML. Illusory correlation, maybe? Or am I missing something?

Posted by Michael Makunas on 2008-01-04

I think you're badly misusing the google trends data. The rubyonrails.org site traffic is also not worth much - after seeing the site once, there's not much more use for it As of early 2007 (the latest figures they've published) O'Reilly was still showing fast growth in the Rails book market. The Ruby and Rails conferences have been rapidly growing in size and corporate sponsorship, Rails, many big-names have been using Ruby/Rails(hulu/oracle/yahoo/sun/ebay), ThoughtWorks now says 40% of their new business is Ruby work, etc.

Posted by crayz on 2008-01-04

I'm not particularly convinced that these searches (or Google trends) are effective methods for noting real ups and downs. Rails Java Does that mean that the "hype machine" was more focused on Java? Nope, it probably just indicates that the Radar site talks about the book market (see the quarterly reports, which often note the effect of Rails on the sales numbers), O'Reilly books (there have been more Ruby-related ones recently, and there's usually a line about the success of Rails), and startups (which often use Rails, which might warrant a mention in an otherwise Rails-free post)

Posted by Keith Fahlgren on 2008-01-04

Michael,

Maybe if you compare similar things, you'll see what I'm talking about.

Trends for Django, Ruby on Rails, Grails and the Zend Framework. Rails is the only one with a negative slope.

Posted by Joe on 2008-01-04

I know it's already peaked for me. I'm on to Merb, Python and ilk.

Posted by Keith Veleba on 2008-01-04

Keith,

Thanks for inadvertently making my point for me.

Yes, in 2005, when very few people knew about Ruby or Rails the Radar site pumped out 40 posts on Rails and a total of 51 posts that mention either Ruby or Rails. That's one post a week. That's hype.

Posted by Joe on 2008-01-04

The TIOBE community index of programming language 'popularity', also shows Ruby having peaked (early 2007 on its graph - but relatively flat after rapid growth from early 2006): http://www.tiobe.com/index.htm?tiobe_index

Michael Foord

Posted by Michael Foord on 2008-01-04

Google Trends doesn't seem like the right measure of real "growth" of Rails, as pointed about by @Michael Makunas above. More likely the Trends data show that the number of people who heard about Rails, didn't know what it was, and Googled those terms has peaked.

Posted by Mike Champion on 2008-01-04

Joe, you cannot measure django in google trends that way, because there is musician with the same name. Also for java, you have: Powerful earthquake rocks Indonesia's Java island...

Posted by Petar on 2008-01-04

REST hype is what bothers me.

Posted by Bill de hOra on 2008-01-04

Mike,

More likely the Trends data show that the number of people who heard about Rails, didn't know what it was, and Googled those terms has peaked.

Right, silly old me measuring the growth of Rails by 'newcomers'.

Posted by Joe on 2008-01-04

Crayz,

That link to the O'Reilly book market summary is interesting. I noticed you didn't point to the most recent one, which seems to show Ruby book sales growth the same as Python sales growth (the little squares are the same color), and sitting at about half the volume (the Ruby square is half the size). The other thing is that between January and May the focus of the analysis was switched from Ruby books to Rails books. It will interesting to see where these numbers are if/when the next state of the computer book market report is done.

Posted by Joe on 2008-01-04

Michael, If you look at the news tab at the right of the google trends page you will see that "HTML" and "java" are vague terms for a search that will include things like the tsunami of 2006. The RoR query, on the other hand, seems correct.

Posted by Ignacio on 2008-01-05

Joe, the search you linked includes comments posted by readers. I'm not sure they should count, as most of them aren't O'Reilly employees or writers. (I also didn't find very many mentions of Rails from 2007.) Statistics are fun.

Posted by chromatic on 2008-01-05

chromatic,

Rael Dornfest, 2005:

Could we be any more explicit about our appreciation for and expectations of Ruby on Rails? Aside from writing about it left, right, and center; including it in our publishing plans; and packing more Rails goodness into our Open Source Convention than you can shake a stick at, that is. (Not to mention Rails hacking in any spare waking moments.)

We've also just put our money where our radar is by supporting Rails Day 2005, a 24 hour Ruby on Rails hackathon put together by Lucas Carlson and friends.

Posted by Joe on 2008-01-05

Joe, April 2005 was 33 months ago. Granted, O'Reilly publishes the occasional book and article on Ruby and Rails, and as far as I know still plans to work with Ruby Central to put on Rails Conf 2008, but if that's all it takes to become a hype machine in your mind, you should see what an organization that really wants to hype something does. Oracle and Sun are particularly good examples. I do certainly believe that the Rails buzz was awfully silly at times, especially in mid-2005, but I'm not aware of any of our coverage since then that's been inappropriate or inaccurate enough that the phrase "hype machine" should apply. Perhaps we understand the term differently though.

Posted by chromatic on 2008-01-05

chromatic,

From the Radar's about page:

At O'Reilly, a big part of our business is paying attention to what's new and interesting in the world of technology. We have a pretty good record at having anticipated some of the big technology developments in recent history.

I think what has been bugging me most is that in the case of Ruby and Rails it seems like O'Reilly crossed the line from 'anticipating' to 'asserting'. As if the rise of Ruby and Rails was predicted and then everything possible was done to make sure that prediction came true.

Posted by Joe on 2008-01-06

Joe,

I can understand that perspective, especially given the book Beyond Java. I thought we did have way too many Rails books planned in Q3 2005, for example (but fortunately we didn't publish all of them).

I've never quite understood what people meant when they said that O'Reilly really pushed Rails as a conspiracy to sell books, especially given that from January 2005 (when we published Curt Hibbs's first big tutorial article) to mid-2006, we published a handful of articles (four or five I believe), Beyond Java, and a few weblogs. To make that conspiracy theory fit the facts of what happened, we'd have to be great at pushing things and lousy at making money off of it.

If we did have an identifiable cheerleading period, I believe it ended in 2006.

I think what you're seeing may be more a reflection of where buzz is among the so-called alpha geeks with weblogs. I'd certainly like to see a lot less technofetishism and taillight chasing there, but to some degree all conversations are circular, so it's not entirely avoidable.

Posted by chromatic on 2008-01-06

Joe - It looks like we linked to the same page, actually. If you look there's "Ruby" on the left in the "sys & prog" section looking around half the size and the same growth rate as "Python". However there's also a separate "Rails" in the "web des & dev" section, significantly larger than the Python one in the other area(I see no python area among web design books), and growing at 202% - with only a couple other topics growing at similar rates among any of the books

Posted by crayz on 2008-01-07

2008-01-04