The everyday blog of Richard Bartle.

RSS feeds: v0.91; v1.0 (RDF); v2.0; Atom.

Previous entry. Next entry.

12:27pm on Wednesday, 8th February, 2012:



Much of the discussion at the conference I was at concerned Search-Engine Optimisation, or SEO. Basically, the problem that developers of social and casual games have is that there are half a million apps on the Apple iStore and getting people to notice yours is hard and expensive. If you don't have a game in the top 10, your products have to be discovered. This means you either have to pay to have them discovered (ie. advertise) or optimise the description so that people searching for a particular kind of game will find yours before they find someone else's. Otherwise, you could have the proverbial best game in the world but no-one will play it because no-one knows about it.

Some platform owners are aware of the problem, and have tried various solutions. For example, one thing you can do is rate games then tell people the most highly-rated games in their chosen category first. This is open to abuse, though, because you can use sock puppet accounts to vote for your own products to inflate their ratings (and deflate your rivals' ratings); this is especially a problem if you order by the average rating given to a game, because games with a small number of (possibly fake) player votes can shoot to the top of the list. If you take the total number of votes into account, it tends to select those games that are being played a lot anyway as they have the most people voting for them. You may as well use nunber of downloads as your ordering system.

There are some other ways being tried. One of the most popular is founded on the idea that if your friends like something, maybe you will? This is the aim of the "Like" and "+1" buttons you see on web sites. For games, if a friend finds a game they like and give it a positive vote, that game will show up sooner in your own search. This is all well and good except it could be that 10 of your friends tried the game but only 1 liked it; nevertheless, without a way for the other 9 to say they disliked it, it's still going to get a high rating. Moreover, it may not be the case that your friends do like the same things that you do: I wouldn't want my wife's taste in games to influence the results of my own search for games, for example.

There is actually a way round this. I've mentioned it before in terms of repuation systems in MMOs, but it's applicable here, too. The basic idea is that when I'm looking for something, I want to get recommendations based on whether I personally will like them or not. I don't want to know how many other people liked it, I don't want to know what other people who liked it liked, I don't want to know whether my friends liked it: I want to know if I'll like it. More generally, I want to know if someone with my taste in (in this case) games would like it.

OK, well we can do that. When you buy or play a game, you get to rate it: like, OK, dislike. You do this with most of the games you play. Your responses are stored in a database. Important: everyone who has the same taste in games as you will give the same ratings as you to the games they've played. If you're looking for a new game that you haven't played but someone who has the same taste as you has played, then whether or not you like it can be guessed at by seeing if they liked it. If they did: it shows up high on the search; if they didn't, it will show up low on the search.

You never know whose opinions are being used. You have no need to know. You don't care if it's some farmer in Iowa or some prisoner in Pentonville or some fashionista in Paris. All you want to know is whether people who have the same taste in games as you liked this game you haven't played not.

This isn't a system open to abuse. If you have a bunch of sock puppets saying they liked a game, their views will only be taken into account if they have a history of other games you liked. Their future opinions will be worthless to you if you try the game and decide you don't like it: that clash of views will mean the sock puppets will no longer have the same opinions as you, so what they say won't be taken into account for your future searches. They only get to fool you once.

There's also an incentive for people to vote on games themselves (and do so truthfully). The more games you have expressed an opinion on, the greater the chance your views intersect with those of someone else so the better the results will be for you.

There is a possibility that your tastes are so refined that no-one else shares them, in which case close matches may have to be taken into account (you said OK, they said like: treat as a match). Even this may not be enough. Ultimately, you may be forced to fall back on one of the existing methods for guessing whether you'll like a game or not.

With this kind of approach, if you really do have a game that's lots of people would want to play if onluy they found it, it only takes one person to find it and like it for it to be recommended to everyone with the same tastes as that player. It's a form of search optimisation performed by the users themselves. They get meaningful results as a result; you get to stand or fall by the quality of your work. It's not nevessarily great news for comnpanies whose games are in the top 10 so get their products seen through other means, but at most there are only 10 such companies.

It works for games, music, movies — anything that requires a judgement of taste. Will someone go and implement it please? Oh, except Apple: they'd patent the idea...

Latest entries.

Archived entries.

About this blog.

Copyright © 2012 Richard Bartle (richard@mud.co.uk).