Archive for December, 2006
I’ve recently announced that I am teaming up with V4 Solutions to build Temple of Osiris as well as another game as a cooperative effort. I’ve also noted that both Gavin and I are still going to be working on non-gaming product independently.
I have already begun working on the next non-gaming product for Antair, and since the requirements have now been hashed out and committed, and development has entered the coding stage (as of very, very late last night), I feel good about announcing it today.
The Initial Idea
I had absolutely no intention of writing a spam filter (or anything else for that matter) for the Blackberry. I’ve been receiving close to 1000 spam messages a day on my Blackberry for months, and all attempts to find a solution, in the form of existing software, proved futile. Of course, there are hacky suggestions galore, including modifying the registry to delay mail relay so that server based spam filters can capture spam before it is delivered to the handheld, modifying certain settings in Outlook, and signing up for a proxy service through which email is routed and filtered before sending it on to your account — but none of these things were what I was looking for. My server captures about 40% of incoming spam, with Outlook successfully hunting down another 15 percent of the leftovers, but because of the way Blackberry mail routing is configured for personal (and some corporate) accounts, the handheld is blasted with almost all incoming spam messages. I could not find a client-side spam filter for the Blackberry … so I am going to write one myself.
I wasn’t going to release the spam filter to the public. It was certainly not going to be an Antair product. But after speaking to a few colleagues who own the handhelds (including those working with corporate units on Wall Street), it became evident that the problem is rather widespread — and everyone seemed to be just putting up with it.
Out of the two dozen or so people with whom I’ve spoken to concerning the idea of building a client-side spam filter for the Blackberry, every single one asked me to immediately let them know when it will be available for purchase so that they can pick one up. A few even requested to be sent the earliest working beta version as soon as it becomes ready. During a recent night out for drinks, John put a lot of thought into it as well, and confirmed that it sounded like a good opportunity to pursue from a Micro-ISV perspective.
Doing a bit of research, I’ve found the Blackberry user forums to be full of people asking for solutions to the spam problem, with nothing but the previously mentioned hacky workarounds given as advice. I think a good market exists for a solid client-side spam filter for the Blackberry.
Of course, no-one can go two minutes into researching spam filters without coming across Paul Graham. I was aware of Graham’s work on the subject for years, and have built Bayesian classification systems before, so I knew that if I was going to build a spam filter for the Blackberry, it was going to use the Bayesian approach. However, while the approach to the problem will follow the basic outline set in Graham’s essay, because of certain system limitations on the Blackberry platform (mainly memory constraints), Antair’s spam filter for the Blackberry will have to take a few turns off the beaten path of a typical Bayesean filter.
Thoughts on Developing for the Blackberry Platform
There are costs involved in developing for the Blackberry. In order to use certain parts of the Blackberry API, a code signature must be purchased from RIM. Of course, Murphy’s Law dictates that all of the API functions I need for this project fall into this category.
In addition, the facilities provided by the Blackberry API are rather limiting — which probably explains both the lack of an existing client side spam filter for that platform, and the overall sparseness of software for the Blackberry. One day into development, I’ve already come across a serious issue with the API which forces me to make some sacrifices with the project. Suffice it to say, I’ve already had words with the API developers at RIM. I’m *sure* that with the amount of weight Antair carries in the world, they took my suggestions very seriously and will begin work on building a custom API for me and me alone.
In the process, a technological barrier may turn up which will make the creation of this thing impossible. But for now, we’re rolling.
After some discussion of mutual interests, Gavin and I have decided to partake in a joint venture between V4 Solutions and Antair. Antair’s current game in development, Temple of Osiris, as well as the next game (yet un-announced) will be developed and marketed cooperatively between the two companies. We will still be working independently on our current and upcoming non-gaming software.
It should be fun…stay tuned.