Shadow Rush’s First Stream in the Wild: Takeaways and Design Changes

Was pretty stoked to find a stream of the game out in the wild, as I have not reached out to any streamers yet. It's 2 hours worth of footage and he tries out all 6 characters, so it is a pretty good overview of the game in action.

View store page While the Steam demo is not yet released, you can play an identical web browser version here: https://ghostlore.itch.io/shadowrush

The playthrough was super valuable as it allowed me to see what kind of issues players might face with the game. My main takeaway is that unlucky streaks feel pretty bad.

Should Shadow Rush solve for bad RNG?

Shadow Rush randomly awards items known as Power Shards every time the player levels up. Equipped Power Shards trigger special powers depending on the player's Attack Phase. There are 40 base powers and 160 Power Shards total: Each base power has 4 Power Shard variants, i,ii,iii, and iv, that correspond to the player's Attack Phase.

From levels 1 - 25, each class randomly draws Power Shards from a specific pool of:

• 6x4=24 Power Shards that represent a class's 6 'signature skills', exclusive to each class until level 25, that convey thematically what that class is 'about'.

• 1x4=4 of X-strike, a general-purpose Power Shard that does magic damage.

• 3x4=12 more non class-specific Power Shards with restorative abilities that recover health/mana and cure poison.

After level 25, all classes draw from a common pool that contains all Power Shards with no class restrictions.

The Problem:

As seen in the playthrough, Solely Singleton started with the Gunslinger, a character designed with a 'buff-stacking' design philosophy in mind. As he leveled up, out of the Gunslinger's 6 signature skills, he got:

• HP Steal

• Expose Weakness

• HP Steal

• Expose Weakness

Leading him to conclude that perhaps these were the only skills in the game, and the rest were still in development.

After that, he began cycling through the different characters, to try and see if there was more to the game. For his trouble, he got:

• Health Potion (One of the non-class-specific healing Power Shards that spawns a health potion at HP intervals)

• Discharge(A class-specific Power Shard so that was something)

• Health Potion again but on a different character

This is basically the Spotify problem. From the article:

For years, listeners have debated whether Shuffle is “truly random,” often noticing patterns or repeated tracks that didn’t feel random enough.

For the last 5 years, Spotify's Shuffle was exactly that – random. We relied on a standard, publicly used randomization method to generate playlist orders that were mathematically sound. But, as we learned from user feedback, statistical randomness doesn’t always translate into perceived randomness.

Think of flipping a coin and getting five heads in a row. It’s a completely valid random outcome, but it doesn’t feel random. That disconnect between probability and perception was at the heart of the challenge we set out to solve.

When we looked at listener feedback, one theme stood out: users wanted Shuffle to feel more varied and less repetitive. Certain songs or artists seemed to surface again and again, while others felt buried.

From an engineering perspective, this presented a paradox. Randomness by definition doesn’t guarantee even distribution, but human expectations do. The task wasn’t to make Shuffle less random, but to make it feel more fair and more fresh.

It is also common to solve for bad randomness in bigger AAA games. For example, Blizzard has revealed that they use 'pity timers' in Diablo to keep track of how long a player has been playing without getting a legendary, and if they go on for too long, the chances begin increasing up to the point they get a legendary item, at which point the counter is reset. And don't get me started on what they do in mobile gacha games...

The Solution:

At first, I thought maybe I should implement a 'no repeats' rule which prevents the game from giving 2 of the same Shards in a row. But as I mulled over it on a Sunday, I realized that this was an incorrect framing of the problem. Because unlike Spotify, randomness isn't completely bad in Shadow Rush, even though the player might feel pretty bummed about getting unlucky with their rolls.

A bad streak nudges players to think of alternatives, like when Solely Singleton decided to try different characters. He also wondered if he could use the shared item storage system to transfer the Shards he had already earned up to that point over to his new character, so that starting over would be faster and easier. This feature is locked in the demo, but this was exactly the sort of reaction I was hoping to see when I made that system.

Shadow Rush is a game where I want the player to make new characters over and over again to experiment with different builds and playstyles. Each new character you make contributes to your overall stockpile of Power Shards and opens up more options to mix/match Shards to create hybrid builds.

The problem was not that Shadow Rush has bad RNG, because from a design perspective, that little bit of friction, and the resulting player epiphany, was a GOOD outcome.

The problem was that Shadow Rush has a bad first-time-user-experience (FTUE).

There was no need to design a complicated system of checks to keep track of what a player recently 'got'. The solution I went with was simple:

• For the first 3 levels, each character class will draw from a bunch of smaller item pools, mutually exclusive so there is no chance of repeat.

• They will consist solely of that class's signature abilities.

• Each pool is essentially an 'either-or' of two families of Power Shards: Level 1 pool contains only Power Shards A (i - iv) and Power Shards B (i - iv), Level 2 pool contains C (i - iv) and D (i - iv) and so on.

This ensures that, by the time the player reaches level 4, they should have a varied experience of 3/6 of the character's signature abilities, and a first impression unspoiled by 'bad' RNG.

This system still has an edge case where, a player who is unaware of this system might start a bunch of the same character class to quickly cross level 1 with the expectation to 'see' all the possible outcomes early on, only to find themselves getting one of two outcomes. But I'm counting on players doing this to be already savvy veteran gamers who would understand from past experience that the game is 'stacking the deck', as games sometimes do. We'll see.

Now that I have decided that this is really an FTUE problem and not a randomness problem, I made several more changes to improve the FTUE. From the video, and seconded by my friend Dorkster (developer of the open-source FLARE engine which Shadow Rush runs on, check it here: https://dorkster.itch.io/) I added some basic item descriptions.

Also based on comments in the video, I adjusted the xp table so that the first level is faster to reach. I also made every character start with 300 gold so that they can buy the first upgrade easier.

The point of this post is just so you know that I am listening and I take feedback seriously, so if you have any thoughts on the game please let me know in comments or reach out to me at my Discord (https://discord.gg/mRJxnEWu). I also need to get around to contacting streamers, but I have not yet made any serious attempts to do so, partly because I am still working on content for the game and partly because I do get social anxiety when reaching out to strangers. So if you know a streamer, or are yourself a streamer, please consider streaming/recommending the game. I will try and get the steam demo version up soon once I iron out more stuff.

Finally, big thanks to Solely Singleton for being the first person to stream this game, his Shadow Rush video is currently just shy of 1k views so please help him reach that. He is a small creator like myself so please check out his Patreon page as well:

https://www.patreon.com/SolelySingleton