Inside The Oliver Twins' user-generated, voice-controlled, make-your-own-game platform

The democratisation of video games development has opened up the industry over the past decade, thanks to the likes of Unity and Epic Games making their engines more accessible in terms of pricing and usability. But there are some that still believe the barriers to game creation can be lowered even further.

Among them are veteran UK developers Philip and Andrew Oliver; better known as the Oliver Twins, creators of retro platforming icon Dizzy the Egg. For the past two years, the duo has been running a consultancy business, Game Dragons, in order to help other developers bring their ideas to fruition. Doing so perhaps inevitably sparked new ideas between the twins, drawing them back into the world of development.

"We've been in the business a long time," Philip Oliver tells GamesIndustry.biz. "We've done things wrong, we've done things right. The consultancy was to try to advise others on what the pitfalls are and what the right things to do are.

"We look at what [Episodes and Choices] are doing and we can blow them out of the water"

"One of the main pieces of advice was to do something new, creative and different. Do not ever go head-to-head with incumbents because you'll lose. They're already in place, already profitable and already have the customers, so how are you supposed to compete? When we were giving that advice, it kept making us think, 'Well, what is there that's new and different, and that there's going to be demand for?'"

For a few years, the pair has been toying with the idea of voice as an input method for games. Building on this as a concept, the pair teamed up with Viewpoint Games co-founder Neil Campbell to form Panivox and has today unveiled their new platform: RichCast.

RichCast is a program for PC and Mac, with a mobile version on the way, that enables users to play through interactive stories, similar to visual novels or perhaps more closely comparable to mobile hits like Episodes and Choices. Now in Early Access, players are not only able to explore the first wave of narrative experiences but also create their own with RichCast Studio, the integrated suite of tools used to build content for the platform.

With the likes of Episodes and Choices already raking in millions of dollars per month, it's a tough market to break into, but Oliver believes that the inclusion of Studio -- enabling every user to become a creator -- will be one of RichCast's key differentiating factors.

"We did our research and discovered what a massive market [games like Episodes represent]," he says. "But then we look at what they're doing and what we can do, and we just go 'Well, we can blow them out of the water'."

The brothers even explored how to create content for Episodes and similar titles, which they say always involves learning a scripting language. Contrast that to RichCast's drag-and-drop style interface and Panivox is confident the barriers to entry are almost non-existent.

"There is no scripting in ours at all," says Oliver. "It's like we have completely avoided scripting, you just drag tiles around. It's child's play. In fact, we actually believe we may have invented the simplest programming language ever, because you are kind of programming but you're just moving tiles around."

"We may have invented the simplest programming language eve. You are kind of programming but you're just moving tiles around"

He adds: "This is not for programmers. Programmers go and use Unity, Unreal, C++ and things like that. We wanted to create an app that targeted artists and creative writers, and was as easy-to-use as the software they're used to, like Word or Google Docs. We wanted it to be easy to create these experiences."

Another key differentiator is the use of voice control. Oliver tells us this was inspired by the fact that while you can talk to Apple's voice-activated assistant Siri, for example, it "never really has anything to say back to you."

"Voice is an amazing input method -- it's just that people haven't actually used it properly to understand and give you decent responses within a creative medium," he says. "At the moment, Alexa, Google Assistant and Siri do understand you and they can give you facts from the internet, but we think it can be used for entertainment... And the best thing to put it into is narrative."

Of course, voice-activated games have been tried before; Amazon's Alexa has plenty, including a series of Pac-Man stories and a version of Skyrim that gives you randomly generated quests to endlessly work through. Given these past attempts, why have voice activated games not taken off before?

"Because the content is lacking," says Oliver. "They have all come up with a basic idea. The problem is there's a lack of content... To get that great content, you have to give the tools to the users. And the tools need to be really easy to use, and really, really powerful. That's the bit that's lacking with the rest of the industry at the moment, that's actually holding it back. Until there is some great content, nobody cares that you can do [voice control]. Because it's not funny, but entertaining. And getting great content is not going to be particularly easy, which is one of the reasons why we're doing Early Access."

One of the other issues with voice-based games is that voiceovers are really expensive to hire, and leave no room for iteration; once you've recorded the lines, it can be costly to make even the slightest tweaks because you need to hire the actors again.

Panivox's solution is AI. There are around 200 AI voices of different styles and characters implemented into the app. Creators simply type the lines, assign it to a voice, and they get the performance, which is automatically adjusted as you edit the script. Naturally, there will be on-screen buttons to press if players are unable or disinclined to use voice commands; with a mobile version on the way in early 2022, Oliver points out some people may not want to shout "just knife him" (for example) in public. But he maintains it will open up gaming to those who do prefer voice controls, and those who might want to play in different situations.

"The fact that it's voice-controlled means that you could just be walking the dog and be playing games, whereas that's not a possibility right now with any current tech."

RichCast experiences will also have visual elements, whether it's imagery or videos the creators upload themselves or something drawn from the growing library of assets. There's even talk of sprites, 3D models and 3D environments in future expansions. But key to a creation's success will be the interactivity of its narrative.

Unfortunately, herein lies RichCast's biggest challenge: writing non-linear narratives that give players meaningful agency is hard. Even some of the industry's biggest and best studios have struggled with this: just cast your minds back to the backlash of Mass Effect 3's endings which, after three games' worth of choices, boiled down to just three similar outcomes.

"The fact it's voice-controlled means you could be walking the dog and playing games. That's not possible with any current tech"

"One of the issues we have is people don't have the skill to use this," Oliver acknowledges. "When I pitched the concept to [Fighting Fantasy co-creator] Ian Livingstone, he actually warned us about that. But this is a new skill people should learn, and we have made it phenomenally simple to learn, and to play with it and test it.

"It's a bit like if you think you're a great actor, let someone film you and watch it back -- you'll find out you're not. But the iteration process is so fast, people will work that out quickly. We're expecting a lot of people to dive in, try out [their stories] on a few mates, and work out very quickly what works and what doesn't. The initial demos we have in RichCast right now are alright, but in a year's time we're going to look at them and cringe, because people will have got so much better so quickly at writing this stuff."

It's not just the skill of creating interactive narratives; RichCast writers will need to learn how to predict what the player may say and guide them to using the correct voice commands. Given the natural language input means players can phrase these commands in a plethora of ways, you'll need to encourage them to use the right terminology your game expects.

Oliver offers an example from a Sherlock Holmes game already available on RichCast. Watson asks the player whether they want to examine the body further or go to the chemist. The keywords here are 'body' and 'chemist', and these options will appear on-screen as prompts, but players might respond: "Let's look at the dead guy."

"Now we need to look for other words that might suggest 'body'," Oliver explains. "So it might be 'dead', 'die', 'corpse', and the system has a method of coping with all of that. It's what the voice industry calls 'intent words.' You had the intention to mean this, but you said something else."

Oliver stresses the importance of playtesting creations on friends in order to iterate and discover other 'intent words'; perhaps someone says "victim," so you need to add that into your script. There is a skill in priming players to say the right words, but it's one Oliver believes some people "will absolutely nail."

"One of the things we found when we were doing SkySaga -- but anybody will tell you this when you do a user-generated content platform -- some people blow your mind as to what they create," he says. "We know that people are going to do stuff [that's] going to really surprise us."

This brings up another issue RichCast may face. As with any user-generated content platform, there are moderation concerns. Oliver clarifies that no creator will be able to instantly make their story available to the masses: first, they need approval from multiple friends in order to request publishing. All requests will then go through a moderation process to check for offensive content, and once published a rating system should help good experiences rise to the top while poorer ones fall away.

Of course, any fiction platform brings up another concern: copyright. Sites where writers can publish their own work inevitably get flooded with fan fiction, and while Sherlock Holmes is public domain, the likes of Harry Potter and Twilight are not. Oliver says this is "one for the lawyers" but adds that Minecraft partly rose to fame from people sharing their recreations of fictional worlds. Every user will have to confirm the story is of their own creation before it is published, but if copyright holders issue takedown notices, Panivox will of course remove them.

"When I said Roblox, I cringed a little because the rules they have for cashing out are so horrible that we will not be mimicking that"

Finally, there's perhaps the biggest incentive for users to embrace the platform: monetisation. While not available at launch, Oliver confirms the plan is that Panivox will eventually enable creators to sell their stories, and possibly even include in-game purchases.

Oliver says the closest model to RichCast is that of Roblox, with players and creators dealing in virtual currency that can be cashed out to real money. Three different membership levels are planned: Advocates, Professionals and Business. The first two will earn a different cut of the royalties from their sales (a higher cut for Professionals), with Business users earning the majority of proceeds from their transactions.

Of course, Roblox has recently come under fire for its monetisation model -- something Oliver acknowledges when we point this out.

"When I said Roblox, I cringed a little because the rules they have for cashing out are so horrible that we will not be mimicking that," he says. "We're trying to be the good guys, we're trying to make sure content creators have amazing tools to create amazing experiences and receive a lot of the money that is generated from it. Obviously, we've got to put a business case together, but we want to reward those people and make sure as much as that filters through."

Creators will also be given a price range within which they can sell experiences, with Oliver adding the emphasis is on making purchases as cheap as possible.

"Our thinking is that if someone is going to spend 15-20 minutes playing an experience, they won't mind paying 10p or 20p for it -- especially if they know this is a community platform. If you don't put money in, you won't get people producing good stories."

And while the obvious application for RichCast might be narrative games, Oliver is keen to position it as something with far broader potential. As Roblox does, he sometimes refers to user creations as 'experiences' rather than 'games', observing that it could be used for non-fiction, trivia challenges, training programs and tours of museums or city landmarks. There are even plans to introduce geolocation input for the latter.

"We don't want to say it's games," he concludes, "because that immediately stops people thinking about what it could be."