Time to Ditch the Dates

When open source software projects establish a fixed development cycle and publish dates on which releases will be made, they are asking for trouble. Many users of FOSS have little understanding or interest in how it is developed. So, when a project states that xyz version is planned for release on abc date, users take this to mean that they can expect to see the new version available on that date.

Many projects manage user expectation by being clear about what users can expect - a group of unpaid volunteers will work hard to get a new release out as soon as possible, but it "will be ready when it is ready". Joomla! and Mambo, among the most popular FOSS CMS's around have often been criticised for this approach, but the simple fact remains - when you have software being developed by unpaid volunteers who need to fit their coding around their work and life commitments, it is impossible to set fixed deadlines.

WordPress is a little different. Not only has WordPress been setting scheduled dates for releases but the fact that WordPress is developed as a project by Automattic, a commercial business, means users tend to rely on the dates. And perhaps they shouldn't.

With the release of WordPress 2.1 on January 22, 2007, Matt stated:

"More exciting for most of our users, though, is our new development cycle. Based on everything we’ve learned in the past 3 years of doing WordPress, we’ve decided to shift to a more frequent release schedule like Ubuntu, with major releases coming several times a year. So, for the first time in WordPress’ history, I have an answer to when the next version is coming out: April 23rd."

Soon after, the WordPress roadmap displayed the notice that "After the 2.1 release, we decided to adopt a regular release schedule every 3-4 months with the features primarily driven by ideas voted on by our users." Versions and their planned dates for release were published.

On the face of it, this is not a bad thing. Announcing dates gives users certainty doesn't it? Well, not in WordPress' case. From the time of this announcement until July 2007, the roadmap showed:

• 2.2 - April 23, 2007
• 2.3 - August 20, 2007
• 2.4 - November 26, 2007

Compare this to the actual release dates: 2.2 released May 16, 2007 and 2.3 released September 24, 2007.

In fairness, Matt did make an announcement on his own blog, on April 18, 2007, to alert people to the delay in releasing WP2.2:

"The WP dev team has decided to hold back version 2.2 for at least a week or two from the original date of April 23 while we polish things up. I’ll post an updated release date as soon as we figure out how long everything is going to take. (Which is extra-hard in open source development.)"

WP 2.2 was released just inside 4 months after 2.1 but even so, the projected date had well and truly gone with this, the first release under the new development schedule.

At the end of January 2008, the roadmap showed this:

• 2.4 - January 24, 2008
• 2.5 - April 1, 2008
• 2.6 - July 3, 2008

Two weeks later, following the announcement that 2.4 would be skipped altogether, the updated roadmap showed:

• 2.5 - March 10, 2008
• 2.6 - July 7, 2008
• 2.7 - November 10, 2008

Great! WP 2.5 was on target to be released three weeks earlier than expected! Clients, some of whom were struggling with bugs, were informed that their pain might be over in a month. WordPress blogs were buzzing with the news but when the date passed with no activity the forums started getting long threads from people anxious to know what was happening. WP 2.5 was released on March 29, 2008, still ahead of the original schedule but 19 days after the amended date. WordPress missed a great opportunity to come in ahead of schedule and have a bunch of very happy users.

WordPress 2.6, released July 15, 2008, was a hair over a week late but this wasn't an issue for most of us. Although single-point releases (ie. x.x) are major releases, 2.6 was not significantly different to WP 2.5 and there had been little hype over what was coming.

You will have noticed that WordPress 2.7 was scheduled for November 10th, 2008. The hype over this release was intense, but so were the delays. There were blog posts announcing delays and estimating release dates, but the roadmap was not updated and there were no official announcements on wordpress.org. Because the changes were significant there were many people delaying launching new sites until 2.7 final was available. It was at this time that I really started questioning the wisdom of setting dates for releases.

So, now we are on to WordPress 2.8. This promises to bring in new API's, new widget handling, changes to how JavaScript is handled, and a number of other new features that may potentially break themes and plugins. The roadmap gave these dates:

• 2.8 - February 2, 2009
• 2.9 - May 11, 2009
• 3.0 - August 3, 2009

Then in January, it was changed to of March 9th. On 10th February, WordPress 2.7.1 was released, along with this reference to 2.8:

"partly due to the stability of the 2.7 we’ve decided to push back 2.8 a few more weeks so we can bring in a few more speed optimizations, taxonomy enhancements, and new theme features. We also need to figure out which jazz musician to name the release after."

By March, the release date for 2.8 was given as 1st April. Currently, the roadmap shows April 27, 2009.

At the moment, WordPress 2.8 shows 469 open tickets. It is still unknown if there will be a beta release or release candidates. WordPress may surprise me and bring in a stable release in three weeks but I doubt it. There is still a lot to be done with the tickets that are set as critical or major, and with so much work still outstanding it's hard to test the impact of 2.8 on compatibility.

This development experiment with setting scheduled release dates is a failure. Each significant release, such as 2.5, 2.7, and now 2.8, has created a lot of excitement, a lot of hype, and raised user expectations to a level where patience is thin. Those of us who work in open source development understand the challenges and can appreciate the hard work that goes into any new release but users often don't.

WordPress needs to rethink its development cycle and stop attaching dates to the roadmap. Let the community appreciate what you have done instead of giving them grounds for complaint over what is not done in time.

What do you think? Have the published dates caused problems in communicating delays to your clients or have you delayed updating or starting a new blog because you have expected a new release on a certain date?

If you enjoyed this post, make sure you subscribe to my RSS feed!