Monday, September 15, 2014

EVE's Six Week Release Cycle: A First Look At Results

CCP is now 15 weeks into its plan of releasing content patches every six weeks instead of every six months for EVE Online that CCP Seagull announced at Fanfest.  While EVE's executive producer laid out a technical rationale for the change, I see a couple of other benefits to the increased release tempo.

First, the standard in the MMORPG industry is moving towards faster releases of content.  ArenaNet really began the trend with the success of its two week release schedule for Guild Wars 2.  But setting such a pace is easier said than done. Carbine is the lastest studio to find out, as Wildstar's Senior Game Designer Megan Starks admitted at Gamescom that the studio was abandoning the attempt to release content every month.  But as CCP demonstrated over the past few expansions its ability to release point releases, I'm pretty confident that the game devs will have no trouble keeping up with a six week release cadence.

A steady release of content is important, especially for a subscription game in a genre now filled with free-to-play and buy-to-play games.  The more frequent release schedule also is consistent with Hilmar's past statements that EVE is a service.  Selling the game as a service will make sense to those used to subscribing to entertainment services like Netflix and Spotify.  The consistent updates will help convince people that the monthly fee is worth the price.

Another benefit I see is in marketing.  Let's face facts.  EVE Online is an eleven year old game.  Barring CCP adding a functional version of Walking In Stations or vastly improved PvE content, the game is probably not going to attract a huge wave of new players who are going to jump into an expansion and play for years.  The old PR model of hyping two big expansions and then relying on player events like Asakai and B-R to publicize the game just wasn't working anymore. 

Going to the six week release cycle means a constant stream of dev blogs that the gaming press can write short stories about.  Between that and coverage for regional events such as EVE Vegas, Evesterdam, and EVE Down Under, and CCP can keep EVE mentioned on the gaming sites.  Keeping EVE constantly visible could draw in lapsed players, especially if CCP fixes features that irritated the former subscribers.  At this point in EVE's life, targeting former players is a viable option.

Theory is fine, but what are the results?  In fairness, I think we really need to wait until early December and the Rhea release to judge.  From the launch of Kronos to the launch of Rhea would represent the typical length of a summer expansion.  But since I'm starting to read and listen to analysis and speculation as to how the new deployment schedule is working, I thought I'd pull out some numbers and take a look at the halfway point.

The chart above lists the performance of each summer expansion at both the twelve week mark and at the end of the expansion.  The year 2009 did not have a summer expansion as Apocrypha launched on 10 March 2009.  I added an entry for Kronos to show the performance of the new release cycle this year.  The calculation was performed by taking the average concurrent user mark (ACU) for the week before an expansion launched and comparing it to the ACU number for either the twelfth week after launch or the week before the launch of that year's winter expansion.

The historical data shows that by the twelfth week after the launch of a summer expansion, one could determine whether the expansion would have more or less activity by the final week of the release.  If that trend continues this year, the first five releases of the new release cadence will result in fewer accounts logged into EVE.   But that could change, as Oceanus is scheduled for release on 30 September, with Phoebe following on 4 November.

I was also interested in the trends after the launch of each expansion compared to the new release cadence.  The above graph represents the change in the weekly ACU compared to the ACU recorded in the week before the launch of an expansion.  An interesting fact is that seven weeks after launch, all the summer expansions (not including the 6-week releases) showed a positive change in the ACU.  Even Incarna.  But the strength of small, frequent releases began to show at the nine week mark, as Crius began to outperform Incarna and Hyperion scored better than Odyssey.

Will CCP buck the historical trend and see the ACU in the first week of December exceed the number set at the end of May?  With two more releases, the possibility exists.  But CCP Seagull's announcement at the end of her dev blog on Oceanus could also play a role.  CCP is currently working on changes to null sec.  She linked to a forum post by CCP Fozzie who spelled out CCP's plans:
"Those of you who watched the Fanfest presentations or the recent Alliance Tournament will remember that we have formed a targeted 'Nullsec Working Group' back in April of this year to lead the way towards our next major round of changes to zero security space. This group consists of CCP Bettik, CCP Delegate Zero, CCP Greyscale, CCP Masterplan, CCP Rise, CCP Scarpia, CCP Ytterbium and myself.

"We have been working on re-evaluating the high level goals for nullsec and sovereignty, surveying and learning from the EVE community’s extensive discussions on the issues, and designing and prototyping potential changes to improve nullsec gameplay.

"The working group recently held an extended offsite design and discussion session on September 5th, in which we discussed many of the player-written proposals about Nullsec, clarified our collective position on several issues and made a lot of progress in preparation for the CSM Summit.

"For this upcoming CSM summit we are planning to discuss in detail a set of significant, specific and targeted changes that we hope to release in late 2014, as well as the concepts and prototypes that we are developing for more far-reaching changes in 2015. The CSM has already proven an invaluable resource for bringing us feedback and analysis on the current state of nullsec as well as the community’s desires for the future. We are confident that the multiple nullsec sessions that we have scheduled for this summit will be extremely valuable.

"Our current plan is to bring the late 2014 designs to the wider community for feedback very soon after the summit, independently of the minutes.
So we could see action concerning null sec, even if it is the form of offering feedback, as early as the end of September/beginning of October.  The first changes to null sec are likely to deploy in the Rhea release on 9 December.  Would the promise of change to null sec, and the coverage the associated dev blogs may receive in the gaming press, lure players back to EVE?  We'll know in a few months.


  1. The biggest issue I see with the new cycle is the decreased BANG factor. Previously expansions were built up to and added a whole heap of new stuff which got people talking. With the new release cycle that is lost, and the releases already out seem to be a bit... meh. Crius added a fancy new UI and mixed up manufacture a bit, but it was only half an update. Invention and POS mechanic improvements are yet to be released, leaving it feeling half done. the next one (is it Hyperion? See, I can't even remember the names of these things) may as well have been a standard patch.

    Personally, I think they should have stuck with the 2 expansions a year, but made sure they actually worked first time. A whole array of new features and tweaks without 3 weeks of additional downtimes would be far more exciting than half built content.

  2. You're not exactly a historian are you?

    "Personally, I think they should have stuck with the 2 expansions a year"

    Because that worked so well over the pre-Incarna years?

    "but made sure they actually worked first time"

    I'm fairly sure CCP were trying to do this. I doubt they were wanting to release buggy and underwhelming "expansions" (hard to call Incarna an expansion).

    Given that CCP could not properly sustain 2 expansions a year pre-Incarna, what makes you think it would be a good idea now? It's precisely because of the whole "must release 2 jesus features a year" thing that CCP now have to catch up on 7 years of missed patches (and they're actually doing quite a good job imho).

  3. I'm not saying they could properly sustain them, but they so far haven't properly sustained the 6 week ones either. The same release issues are rife, the difference being that there's not so much hype over these ones. I can't see major gaming sites wanting to make a huge deal every 6 weeks, so they've lost that edge, and for what?

    And what are they doing differently? All they are doing is releasing the same expansions in parts. The industry revamp will be in 3 or 4 pieces of which Crius was the first, undoubtedly having to revisit balance with each feature change. The only thing different is there's going to be no big bang with each release.

    And yes, the 2 expansions a year worked. It got the game this far. You can't pick out a couple of crappy expansions and use them to deny that the system worked.

  4. You don't read sites like Massively a lot, do you. Their writers are always looking for content so they lap up any press release or public statement that game companies make. CCP doesn't need huge surges of publicity. They just need that steady trickle of information indicating that new stuff is coming out to keep the game in the eyes of the consumer. Compared to other companies, CCP has really been poor at doing that in the past.

  5. I can tell you don't work in any sort of software development environment (or at least I hope you don't!). Your perception that 2 expansions a year was working well is very flawed. You appear to not understand the "cost" of doing that.

    I'm talking about the cost of bugs that aren't fixed (and new ones that are introduced). The subsequent cost of trying to fix them (because this cost becomes bigger and the pressure to release the next jesus feature is huge, the bugs are often left unfixed). The cost of developing on top of horrible inflexible code (massively increases costs of integrating expansions with existing features, content & code). The cost of maintaining an ever increasingly fragmented code base. Then there's personnel issues (key people leave, knowledge of that part of code is "lost", further exasperating code maintenance costs and impacting on bug fixes etc.). There's expansion integration issues that become huge, difficulty in creating a cohesive game due to large & distinct expansions... The list of negatives goes on and on and is bloody huge. It would take me hours (days, weeks maybe) to write out a complete list with reasons as to why the old methodology was a terrible thing for EvE.

    Today, EvE actually has a future because thankfully, CCP realised something had to change. I can't tell you which exact dev methodology CCP are now using but I'd assume it's something similar to "Extreme Programming"

    Had they stuck to their old methods (if indeed one can term it as a "method" and not just a "way"), I have little doubt that EvE would now be close to unplayable with little to no cohesion or integration between it's many large and distinct expansions.

    Anyway, I get the feeling you simply don't understand that CCP's whole approach & development process has changed. This:

    "And what are they doing differently? All they are doing is releasing the same expansions in parts."

    tells me my feeling is accurate.

    You are clearly free to continue believing that you are correct, but simply put, you are flat out wrong for the reasons I mentioned above and many many more. Don't take that from just me (20yrs in software dev), go do some research and take it from the greater part of the software development industry.

  6. As a matter of fact, I am a developer. It seems like you think I must suck at it though. A random on the internet called me wrong, I feel so hurt.

    And mate, it DID work. EVE is not a new game, and their 2 releases a year led them through year on year growth. You can't just leap in and say "it didn't work" without providing any evidence of same. Those bugs still occur, so switching up to a shorter release cycle clearly isn't helping. In fact, it now means that they have to have something of merit to deliver every 6 weeks, so arguably they are under even more pressure. Added onto that, they now have to split large features down into independently implementable blocks which is a hard thing to do while maintaining balance. The sov rebalance will probably be a disaster.

    "I have little doubt that EvE would now be close to unplayable with little to no cohesion or integration between it's many large and distinct expansions."
    How? You think that if they stuck to twice yearly expansions, then CCP would be unplayable by now? What exactly would have changed enough by now to kill EVE? Basically you are saying that if Crius and Hyperion hadn't been released, EVE would be unplayable.

    And thanks for letting me know I'm free to hold my own opinions. I plan on doing exactly that. Clearly you have YOUR opinions on the matter which are exactly that they are opinions. You aren't some oracle just because you assume you are.

  7. The problem there is that the trickle of information is only really interesting to those of us that play, and retaining old players isn't EVEs problem. The problem is gaining and retaining new players, many of whom have decided EVE is boring. Getting to them needs big news of big changes to excite and entice them.

    Think about it, what gamers are sitting around reading patch notes for games they have no intent of playing? Small changes to games, new dungeons, levels, characters, etc, those pretty much go by with only the players of the games really taking an interest. An expansion though? Those you hear about.

  8. So you allegedly work as a developer but have no idea whatsoever about agile development methods, why they have become so popular, what's involved, why they work very well in long term / continual development projects and why they often result in a superior and more efficient end product. No clue about the concepts of technical debt and the resulting high maintenance costs, slower reaction times, terrible flexibility, poorer code quality etc. that misplaced or poor (or non-existent!) development processes can induce.

    Please accept my profuse apology. The evidence was overwhelming in pointing to the person behind Lucas Kell as being someone with little or no experience of today's IT industry. The evidence still says that, but I'll just have to take you word for it, that you are in fact a professional developer.

    That you claim the only difference is "CCP releasing the same expansions but in parts" is damning evidence of your ignorance regarding CCP's changes in development processes / your understanding of these processes.

    That you believe CCP's approach changed with Crius a few months back is concrete evidence of your ignorance regarding this. (TL;DR CCP started their internal process changes several years back post-Incarna. The short release cycle is the culmination of those changes.)

    Given your lack of knowledge about this I didn't expect you to take what I said as fact given that it completely destroys your argument. So I asked you to do some research to confirm what I said is accurate and in accordance with the IT industry's view. I even provided you with a link to get you started. Clearly I was asking way too much of you as you preferred to wallow in your ignorance and throw adhom attacks at me instead.

    Ultimately, you are refusing to educate yourself in the subject of this discussion, preferring to throw out adhom attacks instead. To that end, there is little point in entertaining your juvenile manner. Goodbye Lucas Kell.

  9. Wait, what? I'm throwing around adhoms? Coming from the guy who has taken me saying "I think 2 big releases a year generated mroe news than a small release every 6 weeks", and using that to deduce that I must know nothing at all about standard development principles?

    Simply put, yes, agile development can be positive (though is certainly not the be all and end all of development). But the thing is, regardless of how they do their development, the same issues are rife because the core codebase is still ancient legacy code and they've stated outright that won't be changing. It's not like they have got rid of the bugs and release issues, and it's not like prior to the 6 weekly cycle, they were unable to release patches outside of their expansions. The game was already constantly patched with small changes, fixes and balance alterations. The only difference now is they've done away with the publicity boost that comes with expansions.

    And while I don't believe for a second that they started all their changes with Incarna (mainly because I was at fanfest when they explained their motivations for change and spoke to the devs) it's completely totally and utterly irrelevant, since my entire opinion is not even remotely based around how they develop, it's based around *how they release*, which change with Crius.

    But I get it mate, you like to argue which is why you completely ignored what I actually stated and instead started to whine on about the development principles you prefer. *clap clap clap* You discovered wikipedia.