One of my blog posts about the issues that have affected me with Apple’s decision to remove Lords of Midnight from the App store and Marmalade’s decision to pull out of SDK development was referenced in a Blog post on No Time to Play. I started posting a comment there, but it has turned into a blog post.
Felix the author of the news round up goes on to say:
Dear game developers: either buy a perpetual license to your engine, including source code (otherwise it’s useless), or else stick to open source. Failing that, roll your own. The initial convenience of off-the-shelf code is illusory anyway.
I found myself not necessarily agreeing with the comment per se although it does raise issues that developers should be aware of. So I thought I’d counter the statement in order to bring some form of balance. I think it’s important for other developers to understand the decisions taken as to why to use third party SDKs and not roll their own. I don’t believe that there is a hard rule, a right and wrong way to go about your development, but I do think it important for developers to be aware of the implications of those decisions. For me, I got 5 years of a product. I didn’t consider the risk of Marmalade pulling out of the market, no more than I would consider Unity doing so.
Before I start, here is a list of more high profile apps that were developed under the same system.
- Backbreaker
- Cut the Rope
- Call of Duty: World at War: Zombies
- Doodle Jump
- Draw Something
- Godus
- Lara Croft and the Guardian of Light
- Metal Gear Solid Mobile
- Need for Speed: Shift
- Peggle
- Plants vs. Zombies
- SimCity Build It
- Tetris
- Worms
It’s important to understand that The Lords of Midnight was affectively in limbo long before this licence issue raised its head. As an indie game developer not taking his income from the sales of these apps, both LoM and DDR cannot sustain ongoing development – as much as I would like to.
In this case, the decision to use Marmalade, or AirPlay as it was then, was actually made because of its build system. When Mike and I first started working together, he was Windows based, and I was Mac. The Marmalade MKB project allows you to continually create Xcode/Visual Studio projects from the same base allowing you to just pass the MKB project backwards and forwards and not worry about the baggage of any platform specific IDE etc. From that came the additional benefit of all the other platforms.
It’s important to note that as Mike and I first started prototyping and then after Mike’s death I took on the project fully, there would have been no justification in spending the time rolling our own engine to be cross platform. Had we been only targeting one device, it might have made much more sense. The Marmalade SDK is used mainly for the Device Platform harness, the harmonisation of the few low level functionality we needed. The project uses as little of the SDK as it can. Developing a cross platform harness is no trivial task across iOS, Android, Blackberry, Windows Phone, Windows, OS X etc.. and having to keep it all up to date.
At the time, there wasn’t really an option of any other Open Source alternative that gave us what we needed, especially when you consider the cross platform build functionality that we were using.
From me properly starting the project on my own to an iOS and Android release was less than two months of me coding in the evenings while holding down the day job, living away from home. All other platforms followed quickly after that over the next few months. Within 12 months I released the follow up Doomdark’s Revenge as well. I could not have done that on my own and hand roll the SDK. I could not have afforded to by a source code licence on the whim of a project that was likely not to make any money.
From a indie point of view the benefit of that was HUGE – Sales – hard cash! The yearly cost of the licence was easily being covered by the sales, and that has continued to be the case over the five years. A five year sales window that is already much much greater than the original ever had.
There is one important thing to factor into this, and as much as I hate considering it, it’s been the elephant in the room for a long time, and it still remains true; the initial sales of The Lords of Midnight were helped hugely by the fact that people were still aware of the project because of the death of Mike Singleton just two months prior to release. A year later with the release of Doomdark’s Revenge, that was much much less of a factor. If we consider just that one detail in a normal product release cycle – opportunity. Now, I really hope that no other developer finds themselves in the position where the death of someone becomes almost intrinsic to the marketing of your product, but the point is valid, take too long on your product and any form of momentum gained, intentionally or not, could and will be lost.
Now LoM was never about the money, and I’ve been very open about the sales and profits on my twitter feed, but had the game been more financially successful, I might have been able to focus all my time on the project. And that would have been achieved without the time costly overhead of rolling my own PLATFORM engine. I would then have been able to give myself and the project more future proofing, and either roll my own that solved JUST the issues I needed solving, or moved to an open source engine. However, The Lords of Midnight could have been dead within six months, and the time spent rolling my own engine would have been pointless, unless I had a raft of other projects waiting to be developed on the same engine.
So the decision to take on an SDK for quick access to a market made sense. I wouldn’t do it any other way if I had to do it again – although, open source would now be an option.
Now, let’s look at the problem that actually happened. It wasn’t the loss of the licence that became the issue but Apple’s decision to remove the app from the store. Because I had not kept up with releases, even just maintenance ones, latest SDK builds, 64 bit support, etc… I found myself having to quickly try and solve a problem that I had let come to a head. And I had to do that WITHOUT SDK SUPPORT. Just six months earlier and I would have had help, and the problem would have been smoother. Six months later it would have been mute. There would have been nothing I could have done and an already stagnant product would have dropped from the market unnoticed.
For the record, when Marmalade announced their withdrawal from the market, the source code was offered at a price. But I decided that that cost was not justified for a five year old project that had little income and little future anyway. Had I more projects in the market using the system, and more generated income, I think that decision would have been different.
So in closing, my advice to other Indie developers would be..
1. Choose the right tool/SDK for you but really think about why you are making that decision.
2. Consider your time to market, do you have the luxury of time?
3. Keep you toolsets up to date – little and often.
4. Think about the longevity of your app and what the obstacles are to that. How will you remove those obstacles?
5. Maybe decide when to kill your darlings yourself and not let others make the decision for you. 🙂
I think it’s quite mean to tell a starting indie to “roll their own” engine.
We aren’t talking here about a static monolithic piece of hardware like a Gameboy or a Spectrum. We are talking about Android and iOS and PC and tons of other ever changing, devices and configurations.
Not only the opportunity cost. It’s the unforeseen changes introduced by tens of different agents into the underlying system at almost a weekly basis that could one developer entirely focused with just keeping their engine up to the latest bugs and broken stuff that a random engineer at Apple, Microsoft, Google, or whoever else that has their hands on it, might introduce with the next update.
Much less to develop an actual game.
That is for those that are a truly anal, like the guy that made Braid, that cannot feel comfortable with any known programming language, much less with any engine made by anyone other than himself.
For the rest, don’t reinvent the wheel. Most of the time it’s not worth it.
Of course there is always a risk that the chosen engine will close up next month. But instead of going full “Primitive Technology” and develop your own with sticks and strings, I would go middle way and develop the game in a way that the core could be unplugged from the chosen engine and reppluged to a different one with by just developing a different wrapping layer, should that event ever happened.