Data Mining RPG Interests

Anyone who interacts with any service on the internet is subjected to data mining. Social media, search services, and even ad-hoc thoughts sent into the ether help build a model of who you are, people and brands you utilize, and what you find interesting. If are frightened by intrusive models, stop reading now.

Graphs

The majority of modern online services are built on graph databases. Graph DB’s are not simple SQL oriented databases but rather a complex weave of relationships. Google knows what I search for, and who I interact with on G+. Amazon knows what books, and other media, I consume. Twitter would know who I followed and what I read ad-hoc if I used it more. Everything is a relationship of some type. Think of the idea as Mark knows X, Y, Z. He reads materials about A, B, C. Extend to unimaginable depths.

The concept is almost entirely absent in the RPG community. Most specifically, the community is not inclusive to players – not the designers, writers, artists or talking heads (such as myself). Those individuals want to interact with the end consumer but the majority focus on the intermediate consumer — the game master. Even that is done poorly. Far more people run games than talk about them online. Even more play games than choose to discuss them.

Contextual Keys

Information is king. Once a person begins to explore something in the RPG community, it is nearly impossible to easily expound upon the subject. You can search, drill down a few levels, and then end up back at the search engine to find you relevant results. Using technology, that same search could result in relevant blog posts, products, and even discussions on social media. How? Context of the search and the relationship graph. Each and everything reachable on the internet has clues embedded in the text, pictures, and videos. Those cues coupled with the graph model reinforce what is of interest at the moment.

The blog aggregation model is a dated relic. While minimally useful for browsing to find a topic of interest, nothing ties the individual bloggers together. Five people can post on the same topic but unless an author makes an effort to link other posts on the topic, the user browsing the topic hits the end point once it is read.

What if the aggregator had more knowledge? What if it actually extracted contextual keys from the post and could suggest similar articles? The idea is possible. Significant research has been undertaken to convert random bits of text into semantic keys. Those keys, in turn, can be matched to what is being read to suggest other alternatives. Taken a step further, that information could tie into a product database to suggest relevant products.

If it went a step deeper, the hypothetical service would also utilize the information being sought in tandem with that friends and associates have also read. Remember the graph? I’m friends with Bob. Bob read a bunch of zombie articles and recommended a zombie game. At a later date, I’m exploring zombies myself, and Bob’s history pops up suggested articles and the product he purchased.

How would it work?

Two different models need to be considered: producer and the visitor. The producer works for those few who already want to promoted something or expose their ideas to a larger audience. The visitor is the ultimate goal. The mechanism must be useful enough to drive them to disparate sites for readership, purchasers, or just to continue to explore ideas.

Producers have to start the process. They must agree to open consumption of their materials. For bloggers, feeds must be complete, not snippets of articles. To establish the relationship model, the idea of restricting information so an individual will be forced to visit a site must be abandoned. Likewise, the sales end-point for products needs to establish an API that allows contextual scraping. Not the product itself, just the marketing blurb. Opening the relationship information in the form of reviews would also reinforce the interaction. Recognizing that exposing information will result in visitation and perhaps sales should be trivial.

The experience for consumers has to be the predominate goal behind the idea. Bloggers sell articles in terms of page views. Publishers do so in the form of books. The idea has to open doors for the consumer to not only view posts but also provide a chance to buy a product. Consider traps as a topic. Hundreds of buried posts are available about constructing a good trap, thousands exist for specific implementation, and perhaps a handful of products to buy.

If the Consumer, C, was interested in traps, she should be able to find everything related. No matter the heritage or source. If C kept browsing Al’s trap posts, Al’s Book of Infinite Traps, would be highlighted (relationship, interest). If C chose to browse about, every article from A-Z on traps would have equal weight.

Final Thoughts

Nothing suggested is easy. The entire idea is based on personal frustration coupled with a realization that it is possible. While I’m a fan of the OSR, that doesn’t prohibit me from pondering modern ideas. A solution will exist in the near future. Perhaps it will be be someone with an entrepreneurial spirit or one of the big entities will allow specialization. Who knows?