My year as a Totem Pole

It’s been a year since Totem Topple came out, and six months since the patch that fixed it, so feels like a good time to take a step back and analyse what went right and wrong with the game.

The fact that it had to be extensively patched last summer attests to the issues with the original version that launched on PC and Wii U back in November 2015. You can read more about the making of the game, how it faltered, and efforts to rescue it, here.

However, I want to dig into the game design itself. For those unfamiliar with Totem Topple, it’s a tower defence game in which you literally play as a tower, in the form of a Totem Pole. Players select which heads to add to their Totem Pole. Each head on the totem pole has different abilities and stats. Some are turrets, others are heavily armoured or give bonuses to other heads:

I’m a What?

There is an immediate problem even with explaining the game. It’s not obvious who you are actually playing as. Totem Poles are inanimate objects, so it’s difficult to conceptualise being one. Nor do the individual heads have any character or personality or appear to be alive or anthropromorphised in any way. It might make sense if the totem pole was being built by someone, a tribe say, scurrying around at the base. But the game gives no indication that this might be the case.

It’s not clear who or what exactly the enemies are either. Nor what their motivations are for knocking down the Totem Pole. On occasion, some people watching the game’s trailer even thought they would be playing as the enemies, attacking the Totem Pole!

Equally, why do new heads appear from the sky and land on top of the Totem Pole? Usually Totem Poles are made from a single tree trunk, rather than separate blocks.

Furthermore, why do the enemies attack the base? It makes sense from a design point of view as it means enemies have to pass by all the turrets first. Plus the oldest heads are attacked first, meaning players have time to build new complex structures without feeling under attack constantly. Having the top attacked would mean mistakes or choices from early in the game, when the player might not have enough resources for an optimal setup, would linger at the base of the pole.

But having the base attacked instead makes no sense from an outside perspective. Once the head at the base of the Totem Pole is destroyed, players might reasonably expect the tower to tip over as physics (or the game’s name) might suggest. Rather than the whole tower falling down vertically by a single head’s height, as actually happens.

Probably a bit of story and exposition could have helped at least some of these issues. Even then, anyone who skips the story or doesn’t really bother with the narrative would be left somewhat lacking in agency / motivation. (It’s quite common for some players to want to just jump in and get their hands dirty when playing a new game, then worry about the why later).

This is further hindered by the art / theme. Players might be expected to come with a bit of background knowledge from the real world or playing other games. Shields = defence, swords = attack. But in Totem Topple, Bear = ??, Owl = ??. In fact, most of the heads are damage or defence modifiers, with the actual shooting done by the bird beaks attached to the sides.

Say instead, players are building and defending a spaceship, rather than a Totem Pole. Then things are a little easier. A deer becomes a “shield generator” and a bird beak becomes a “plasma turret”. I still stand by the decision to go with the Native American theme. Where the majority of games are set in space or medieval-fantasy land, having something else can help a game really stand out in the market. And I’m glad in the original game-jam where Totem Topple was born, we did manage to find a theme with a strong aesthetic, and that made sense for the core mechanic. In hindsight though, it brought as many problems as it solved.

Tradition

The genre was another area where Totem Topple eschews tradition. It’s ostensibly a tower defence game, but attaching that label brought with it player expectations that simply weren’t matched by the design.

One case in point is the lack of geography in Totem Topple. There is no decision as to where to place the next Totem head. It will always be on the top of the pole. This simplification I feel works quite nicely, as it lessens the sharpness of the learning curve. In many tower defence games, poor placement in the early game can really harm the player. There’s no map to scroll around either. The whole game fits onto a single screen with ease. And there are simply less turrets to worry about. No complex mental calculations of “I have 6 plasma turrets and 5 railgun turrets and 8 beam turrets, so I need an extra flak turret and maybe 2 more power generators”.

That said, learning the nuances of a particular map can be one of the more fun aspects of some tower defence games. And where Totem Topple falls down in the simple / elegant stakes is with the wing and beak placement. It’s not at all intuitive to have the side-parts placed on the next-highest free slot on that side of the Totem Pole. (Even saying that sentence is a bit of a mouthful).

Another example of moving away from tradition is with combat in the game. In a normal tower defence game, turrets auto-aim, shooting enemies when they come into range, before turning to aim towards them until dead or out of range again. Whereas in Totem Topple, they simply shoot a constant stream of arrows horizontally until pointed by the player at a specific enemy.

I actually quite like this concept. By leaving the turrets to their own devices, they provide a screen against enemies that wears them down as they pass. Rather than having strong enemies suck up all the tower’s firepower whilst smaller enemies can just waltz through, as happens with many tower defence games. (As if whoever is manning the turret is completely unable to prioritise. To make an intelligent decision to just stop shooting that bullet sponge for just a second in order to kill the fast, weak, kamikazi enemy that is about to get through).

In Totem Topple, players can choose to target an enemy if they can see one is nearing the bottom of the tower or is taking more damage than just one side of the tower alone can handle. Then whilst making a decision on what to target next, the turrets all go back to auto-fire, helping take out the enemies that were previously getting away unharmed.

Post-Patch (or what went right!)

A number of other elements of the game were reworked and much improved once the game was patched and expanded in the summer of 2016. The enemies could have had a little more variation, but at least they had a series of different behaviours, stats and special abilities to provide a range of challenges to the player. The fire enemies in particular, cause a degree of panic once players realise they can set their Totem Pole on fire! And even more panic once the fire begins to spread.

The water and ice enemies are a little weaker conceptually. The ice enemy merely freezes the Totem Pole, preventing a few turrets near the base from firing for a short period. Whereas the water enemies simply spawn new miniature enemies every few seconds. This at least gives the player some thinking to do when picking what to shoot next.

The game economy in Classic mode is another part of the game design that worked well on the second iteration / post-patch. The game is quite generous with resources in the earlier part of the game, but those resources can quickly seep away if players spend rashly or fail to defend properly. The downside being, when playing with two players on Wii U, that the playing building the Totem Pole can put themselves out of a job for long periods of the game by being too good.

The tutorial I’ve written at length about in the past, but at least it seems to get the job done for the most part. Whilst the jump-through-hoops style isn’t ideal, it does teach players how to get going, even if it’s not great on many of the details or more subtle elements of the game.

TL:DR

For all it’s flaws, Totem Topple has some interesting design ideas, and still comes out as a fun, if slightly confusing game. In many ways it’s similar to Clash Royale, with that game’s simplified tower defence across two lanes. My hope for Totem Topple is that others will learn the lessons from the game and perhaps open up some new thinking when it comes to tower defence genre.

Elements of Damage

Introducing damage types into Totem Topple proved to be a relatively quick way of creating more variety and dynamism within the game. It allowed for new and interesting enemies to emerge later on in the game, and added an extra dimension to the game without burdening the player with too much extra to learn.

ice arrows 1

Change of Plan

Originally, the plan was that whatever damage type was selected when the player placed a beak (turret) or wing, that would be it’s damage type for life. However, this lead to issues whereby players would be stuck with a lot of ineffectual turrets that didn’t match the damage type of the latest wave of enemies. Or worse, not have enough supplies to build more beaks for the incoming enemy damage type, and be unable to collect any more due to being unable to kill those enemies!

Furthermore, it hugely complicated the tower building elements of the game. Since the game doesn’t have an undo function or any way to adjust the tower once built, having to account for different damage types in different places on the tower just put the game that bit beyond the player’s easy understanding. Placing mismatched wings and beaks on either side of a head because you forgot you’d changed damage types gets old really fast.

As well, firing in Totem Topple is done by simply aiming all the turrets at a single target. It doesn’t make much sense firing every damage type at an enemy when the player knows only a few arrows will have an effect. Especially if there are, say, both fire and ice enemies on screen, the player is better off letting the arrows shoot their respective damage types automatically. Rather than half their effective firepower by targeting a single enemy on which half their arrows will be ineffectual. That takes away the fun and point to being able to target specific enemies in the first place.

ice enemies

Limited Tension

Instead, having a single damage type fired by all beaks made both play and implementation that much easier. However, allowing players to constantly switch damage types made things too easy. It took away the challenge of managing damage types, as the player could just spam the button to change damage type to whatever.

The idea of giving the player a limited number of changes in a given time period came about more by accident than design. Copying the other buttons used to build heads/wings/etc meant the change damage type button inherited a cool-down timer. That on its own was too harsh on the player, as often they needed to cycle between a couple of damage types to get to the one they wanted. Giving them a small number of changes, that would then replenish slowly over time, gave just the right balance. Forcing the player to be conservative with their damage type changes, but not to the point of being punishing.

Ready, Aim, Fire!

Some years ago, I participated in a game jam using Eye Tracking technology. My team’s game was a horrible mess, but the eye tracking technology itself was far more responsive and accurate than I anticipated.  When it came time to port Totem Topple over to PC, I decided to have a go at integrating eye tracking, just for fun.

I used the eye tracking to let the player target enemies and shoot all arrows towards that chosen enemy until it was dead. After which the arrows would revert to their usual patterns until a new enemy was targeted. The result was really fun, to the point that people would forget the actual tower-defence and building aspect of the game, especially in Classic mode, and just have a great time shooting as many enemies as they could.

WiiRemote4

Later on, when creating the Wii U updates for Totem Topple, Wii Remotes proved a relatively straight forward substitute in for eye tracking (which Wii U doesn’t support). Pointing the Wii Remote at the TV and letting one player direct fire, whilst a friend or family member concentrates on building the tower and controlling damage type dealt, added a surprise extra social dimension to the game. In Off-TV mode, the enemies are seen on the Wii U GamePad, and the player can tap on the GamePad touchscreen to target them. Equally for the non-eye tracker PC version, players click with the mouse on enemies they wish to shoot.

 

Totem Tutorial the 3rd

(Epilepsy warning: Contains flashing images)

The tutorial system in Totem Topple is something that we’ve wrestled with over a number of iterations. In the launch version of Totem Topple, it worked, in the sense that it gave players enough of an understanding of how the game’s basic mechanics worked to be able to play. Similarly, the complimentary Help system worked well in allowing players to see the underlying stats and numbers for the various heads and enemies.

However, the fact that many players resorted to using the Help showed how much the game was lacking in visual feedback. Many of the minor changes made to Totem Topple for the 2.0 release try to give players a better sense of how their actions affect the game.

Circular Confusion

For example, some heads give a bonus to their neighbours, as indicated by the radius wheel behind them. Unfortunately, the wheel itself, whilst it is clear where it is coming from, it’s not clear what it actually does. There are no visible links or changes on the heads it affects. In fact, some players confused it for some sort of shield or radial attack.

classic_screenshot_01_a

Instead, now indicator arrows point between heads, wings and beaks, allowing players to see when any one element on the totem pole affects another. For health buffs, the affected heads gain a purple glow as a show of high strength, at least give some indication to the player that this head is different to others. However, we’re still looking at ways to make it even more obvious when one head or wing afftects the health of another.

Go Faster

When it comes to rate of fire changes, the differential between the lowest and highest rates was not big enough previously. Whilst the way the game was balanced, a higher rate of fire made a big difference to the player’s ability to kill lots of enemies, unless they looked really closely, players were unlikely to be able to tell that.

To improve this, firstly the balance of the game was changed so that low rate of fire really did mean one measly arrow every couple of seconds, whilst the highest rate creates an almost constant stream. Further to this, a higher rate of fire actually means the arrow physically moves faster. The arrows also gain “go faster stripes” or speed lines as they are called by illustrators. These emphasise the speed, but also give an extra visual pointer to show that things have changed when adding a wing, or that this head-beak combo is better than another.

rof motion stripes 1

Adding in a simple, two step animation of the bird beaks opening and closing every time an arrow fired also helped to give players an extra visual clue that one head/beak is a lot more active than the others. Especially so when arrows are flying in all directions.

beaks open

 

There were also a few combinations where beaks would have a zero rate of fire or arrows would do negative damage. To represent this, beaks would be closed, or fire broken arrows. Even so, this is probably a bit confusing for the player, so long term, we’re aiming to remove the need for these by rebalancing.

broken arrow 2

Flaming Arrows

For damage, arrows now start off black and then slowly turn more and more brightly coloured depending on the damage type being fired. The highest damage dealing arrows are then given spectacular trails of flame, snowflakes or water jets to really drive home that these arrows were specially powerful.

flame arrows 1

Differentiation

Another problem was simply telling heads apart from each other. It’s not intrinsically clear what a bear or a wolf or an eagle does or represents. Unlike, say, a scifi themed game, where the players can work things out using existing knowledge: That gun-looking structure is probably a turret, and the box that emits a glowing shield is probably a defensive mod.

Previously we’d coloured the heads according to function. Blue for defensive and orange for offensive. However, players didn’t pick up on that, viewing each head individually, and trying to work out what each did one at a time.

simplified_enemy_small

So we gave each one a unique colour. We also simplified the designs. A lot of detail was getting lost or creating a pixellated mess on lower resolution screens, which made the heads all blur into one. Simplifying the graphics also helped to make the overall game aesthetic look cleaner. We did the same for the enemies, using the original higher-resolution enemy images for the much larger, newly created boss enemies. This had the added bonus of making them appear more special compared to their smaller, plainer counterparts.

Hitboxes

Another complaint from the launch version of the game was that the hitboxes were too small. Players would see their arrows whisk right through enemies, apparently failing to hit them unless the arrow hit dead centre. However, increasing the enemy hitbox size caused a different problem, whereupon enemies would often hit the second bottom totem head as they moved into the base of the tower. The game is programmed as such to ignore these hits (as it massively complicates the game and its coding if non-bottom heads / heads half-way through the stack can be killed). As a result, the new enemies would often completely fail to knock the totem pole down.

The solution was actually to make the arrow hitboxes significantly larger instead. Some re-balancing in other areas made sure that the game didn’t become too easy as a result.

De-confusing

Some aspects of the game didn’t get a mention in the tutorial, yet were not common features to other games in the genre. For these, rather than let the player figure it out for themselves as before, “interrupt” style tutorial messages would pop up informing the user of key bits of information the first time they did something significant but otherwise non-obvious.

For example, the placing of towers and turrets is a key part of most tower defence games, yet Totem Topple eschews this as one of the limiting factors of the game, forcing players to plan ahead a bit more than they might otherwise need to. That’s all very well except that most players didn’t understand that, especially with placing wings.

nextwing indicators_3

So when a wing is first placed, the game now informs the player both about the functions wings perform and where they will be placed by the game on the totem. Small graphical indicators have also been added to help players subsequently quickly see where the next wing will be placed.

Arbitrary

The other related issue is with the height warning. Way back when the game was first created, it would let players simply keep building up and up ad-infinitum. However, this meant players could for the most part mash the place head buttons (at least in frantic mode), and it didn’t really matter what heads they placed, just how quickly they could throw them down.

To get around this, a mechanic was added in which players were punished for this mode of behaviour, spawning huge waves of super-hard enemies if they built beyond the red flashing warning line near the top of their screen.

punish spawn

This rightly confused the hell out of most players, who did not make the connection between building “too high” and the game suddenly ramping up to impossible difficulty levels. This threat of “angering the spirits” and super-hard enemies did not raise the tension of the game as hoped. Rather, the seemingly arbitrary nature of the players’ demise just as they seemed to be making progress left many frustrated and feeling cheated.

So we simply placed a hard limit on how high players could build. No more than 10 heads, and then the first time they hit that limit, an interrupt would inform them of the limit. The red warning line was kept in, but now as a reminder of that limit rather than as an easily mis-interpreted signal of impending doom.

Sharpening Edges

The game’s UI was also subject to improvements. On the tutorial, enemies would flash and buttons “ping” and scale to help cement the messages in the tutorial text, and better guide players towards the buttons they needed to press or things they needed to be aware of.

ui flashing enemies

Various factors meant we were stuck with using Unity’s less than perfect legacy GUI. However, even simple changes, like using a font other than default Arial also made a big difference. Or adjusting the UI buttons and panels to have angular corners to fit better with the angular background art style.

Overall, the various systems that help the player learn now feel like they do a much better job, working together and in different ways to convey information. There are still many more improvements to be made, but with the changes made, the game doesn’t leave players so lost and confused as it once did.

Rescuing Totem Topple

I was pretty gutted with how Totem Topple was received after it launched back in November, both in terms of poor reviews and poor sales. The first reaction is always to get angry, but looking back on the comments, I quickly realised that most of the criticism was justified. When placing myself in the shoes of reviewers or players coming to the game with fresh eyes, the responses to the game were perfectly rational.

One video of the game was particularly telling: I watched as the player died, died again, died again. Then they checked the in-game help, compared what all the different totem heads did. They devised a strategy, and I’m sitting there thinking “Looks like they’ve got it now. Should be ok this time!”. Only to be taken out by chance enemy spawn that really, they couldn’t do anything to mitigate. The video ended there. Although there was no commentary, I could almost hear them thinking “I did everything the game told me to and still died every single time. **** this.”

The proper, professional response to this is to learn the lessons and bring them to the next game, but otherwise move on. Instead, here I am, still working on the game 6 months later. A mix of sunken cost fallacy and bruised pride sucking me in, leading me to try and rescue Totem Topple. This is the story of how I’ve reached this point.

Rewind

For those not familiar with Totem Topple, the premise is players add different heads, wings or bird beaks to their totem pole. These represent the turrets, walls and stats boosters of more traditional tower defence games. Demon spirit enemies then float down from the top of the screen, attacking the base of the totem pole.

When the bottom head on the pole has taken enough damage, it is destroyed, and the whole tower drops down one level. When all heads are destroyed, it’s game over.

King Jam

Totem Topple, like so many games these days, was born in the fires of a game jam. Over 2 days, myself and 3 fellow indie devs poured our efforts into creating a little mobile tower defence game at a game jam curiously sponsored by King, of Candy Crush Saga fame. After the jam, all the other team members had full time jobs, so I took on the task of polishing up our creation and putting it on Google Play. That took about 2 weeks, and whilst it got a smattering of downloads, it largely went unnoticed, another drop in the ocean app store.

kingjam

I’ve heard a number of theories about why jam games do not make good candidates for full-production games. I personally subscribe to the idea that the best jam games really drill down to one central, super-polished mechanic, but as a result lack depth. Fun for 5 minutes, but without the larger game loops that keep the player engaged on an hour-by-hour basis, or make them spend their lunch break devising strategies for when they get home to the game.

Custom Game

This, plus a nagging feeling that the audience for console games would reject anything that looked too much like a “crappy mobile port” lead me to spend a lot of time creating a complicated “custom game” system. Players could tinker with all the variables behind the scenes, add new heads and enemies, re-order when different waves spawned, and so on.

I was encouraged when a couple of months into development, Nintendo announced Mario Maker. Suddenly, level editors were all the rage and the custom game feature felt like a part of a wider movement around giving players more control to modify and create content in-game.

screenshot 13-02-15 05

In retrospect, I was putting too much time into a feature that very few people were ever likely to actually use. It also allowed me to fool myself when it came to balancing the game, allowing me to say “Oh, well balancing the game is hard, so I’ll throw it open to players and they’ll appreciate being part of the development process, with a developer who is humble enough to admit (by implication) that they don’t know all the answers.”

When really, the game would eventually release without the Custom Game feature, owing to the difficulty of making the user experience feel really smooth. And due to the difficulty to keep it maintained every time something else in the game changed.

As for the balance (or lack thereof), that would turn out to be the biggest issue people had when the game did eventually launch.

Down to Earth

Six months after its inception, the game was “finished”, and I was busy trying to drag the it through Wii U lotcheck process. One of the reasons for bringing the game to Wii U was to go through the process at least once, with what I thought was a relatively small, simple game. Use the lessons learned in helping bring the bigger, more ambitious Flight of Light to the Wii U, whilst also hopefully earning enough money from it to pay some of the artists and musicians that game would need to get it completed.

And I have learned a huge amount about that process, to the point that I’m confident in taking on projects like porting Gear Gauntlet to Wii U, or in simply offering up my experiences and advice to other developers going through the process for the first time.

big indie pitch brighton 2015

However, I didn’t make things easy for myself, insisting on adding odd features like Miiverse integration, leaderboards and analytics packages. Whilst I was wading through all that, I took the game to Pocket Gamer Big Indie Pitch, at which the judges eviscerated the game for its various flaws.

I went away and decided the game needed some big changes. I removed the Custom mode and added in a “Classic” mode to replace it. The theory being that it was closer to more traditional tower defence games, and would be a good way for people to learn the game before going on to the original game, now rebranded “Frantic” mode.

Balance

When the game did come out, Classic mode had had far less time put into its balancing. Whereas most of the reviewers who actually played Frantic mode preferred it, universally, they all struggled with Classic mode. I’d pushed Classic mode to the fore of the game, placing it first on the menu, plus the name itself suggested it was a more logical place to start the game. Consequently a majority of reviewers took Frantic mode, which had far more time put into its balance, as more of a bonus mode, rather than the real core of the game.

Timing

The game also came out at a bad time. Right in the middle of the Christmas/holiday season, a day before two big 1st party Nintendo titles landed on the system, in a period full of the latest AAA games arriving on other platforms.

share-mario-tennis

In terms of marketing, I stood on my virtual soapbox shouting at social media, but got nothing back. I wrote a bunch of blog posts about what I felt were interesting aspects of the game – It’s Navite American Inspirations, it’s day/night cycle, the differences between frantic and classic modes. Few beyond my immediate circle of friends read those posts, and certainly none of the reviewers picked up on the subtle things I’d included in the game, that to me represented the love and attention I’d put into it. Like how the music changed when the enemies started getting harder, then changed back when they got easier again. In fact, one reviewer claimed the game only had one single repetitive song on a loop. Which annoyed me intensely until I realised they probably never got past the first few waves, and so never heard the change.

Just Ship It

The other reason I gave up on making all those blog posts was that in the time between the game getting approved, and actually being available to buy, I’d run out of stuff to talk about. I thought I was being smart, storing up things to talk about with regards to the game. Keeping my power dry until the last couple of weeks before release. Only to find being approved does not mean the game is on the store next week (or anything close to that).

That exhausting process and a feeling like I’d already spent way too long on the game contributed to an attitude in the weeks leading up to the release of “just ship it”, and “so what if it was the wrong time of year”. I was determined to get the game out and into the world, and not be one of the long line of indies who never actually finish a game.

justshipit

Another marketing fail was the video / trailer for the game. It just did not show what it was like to actually play the game, or even whether the player was trying to knock the tower down or build it up or quite what they were supposed to be seeing and doing.

Finally, I’d decided that getting youtubers and streamers to play the game was all the rage, and would make the big difference. So I planned to spend the couple of weeks preceding launch just handing out as many codes to as many people as I could persuade to give the game a try. However, Nintendo only gave me the codes 3 weeks before launch day, giving me just enough time to contact those specialist Nintendo review sites who I knew could make a real difference to the game’s reception.

Once their less than stellar reviews started trickling in, I immediately paused all my marketing plans, and determined that I must find and correct what was wrong with the game before continuing.

Christmas Patch

Another lesson learned the hard way was that patching is its own entire thing, and not simply a case of changing a few variables in Unity and pushing a fresh build to the game via Nintendo. In the heat of the game’s launch and the days after, I made a series of rushed changes to quickly re-balance the game and sent them off to be approved. The resulting patch did not arrive until end of January, long after the game’s fleeting moment in the spotlight of Nintendo fan interest had long since flickered out.

cheese

In fairness, a couple of reviewers did the game a decent service and took a second look, or waited till after the patch before giving their verdicts. But for it to have any impact on the game’s overall reception and sales, that ship had long since sailed.

Onward Totems

That brings me to today. The second patch is now out on Wii U, and the changelog is as long as my arm. New features, new enemies, improved tutorial and UI, and everything rebalanced from the ground up. The game is releasing in new territories (Australia and New Zealand). Taking advantage ofUnity3d’s cross-platform building, the game now runs on PC, and hoping now to pass Steam Greenlight. You can see some of the changes made in the new trailer below:

(You can vote on greenlight here: http://steamcommunity.com/sharedfiles/filedetails/?id=691517098)

Will it rescue Totem Topple? Financially, the chances of even modest success are vanishingly small at this point. However, the game as it is now, at least I can be proud of it, and give those who do take a chance with it a fun time.