The Kirov Metagame

Before poker, and before Counterstrike, it was Red Alert 2. I think my love of Command & Conquer is well-known by this point, but it reached its apex with RA2. Although not the most well-designed mechanically of the franchise and with some fairly severe balance issues, it was the first game in the franchise that brought with it extensive online matchmaking options and a competitive ladder. I dove into it. The 1v1 competitive ladder was “reset” each month (or each three months, I don’t recall perfectly) and I spent a lot of 2003 as the highest ranked player in the world, reaching the #1 spot at various months throughout the year. The game (including the expansion) consisted of three sides – the Allies, who were astoundingly strong, “Yuri” who was probably in abstract terms the most powerful side but difficult to micro correctly, and the Soviets, who were terrible. Both the Allies and Yuri had a range of extremely powerful special units and unique strategic options, whilst the Soviets had no strategic benefits specific to their side and were lumped with a range of powerful but extremely slow units, and no real long-range or support options. However, I played as the Soviets, and the epitome of this was a Soviet unit called the Kirov Airship.

In competitive play, these units were considered a joke. Although possessing the most powerful individual weapon in the game – bombs dropped directly beneath them, able to decimate any building in a few hits – and an impressive amount of health, they were painfully slow and had no way to defend themselves other than dropping bombs straight down to the ground directly below. However, I saw potential in them, and became (to the best of my knowledge) the first person to seriously use these units in high-level competitive play. This is the tale of the strategies I developed to actually get some value from this seemingly-useless unit, and an exploration of an interesting unit that could have significantly more value than its mediocre in-game stats would suggest.


The key to the strategies I developed is the fact that Kirovs have a feature unique to them among all vehicles. This unique feature is similar to how “superweapons” work. Each side has two superweapons which can only be constructed upon reaching the highest tech level. When you construct a superweapon, a timer appears in the bottom corner of the screen, and the fog of war around the superweapon is revealed to all players. This timer counts down from five minutes for the “lesser” superweapons, and ten minutes for the “greater” superweapons, and upon reaching zero – if the superweapon has not been destroyed by an enemy or sold by the player who built it – it can then be used. This immediate warning serves a number of purposes – it fractionally reduces the value of these extremely powerful weapons by warning the enemy about their existence, it gives the enemy a little bit of information about their opponent’s base, and it also serves to amplify and change up the pace of the game. For the player who constructs the superweapon, the overall strategy of their game shifts to defending that structure (unless it is intended as bait); for the player who is faced with an enemy superweapon, their strategy shifts to either destroying it, or constructing their base and placing their units in a manner that will minimize the eventual damage.

However, the game change when a Kirov is built is more subtle. It is not revealed on the map, and there is no timer noting when its attack will be ready (since it can attack as soon as it is built, like any unit). Instead, a sound plays – a voice clip that announces “Kirov Reporting!” to all players. All the enemy player knows is that a single Kirov has been built, and that – although desperately slow and somewhat unwieldy – this unit will single-handedly destroy their base if it is allowed to gain entry. Once the sound has played they know it is out there somewhere, moving to some location on the map, and needs to be dealt with. This voice clip was presumably implemented because the designers felt the Kirov was a sufficiently powerful unit to merit its own sound clip upon completion (something only unique infantry units also possess) and warn other players about what was coming their way. However, alerting players in this way allowed for a range of interesting tactical options and ways to exploit enemy play, all dependent upon the fact that one’s opponent is informed about the creation of the Kirov, but knows nothing of its movements.

Force the construction of AA

The construction of a Kirov necessitates the production of anti-air (AA) units or buildings for your opponent. Even if Kirov’s may be slow and obvious, their destructive power cannot be ignored. A Kirov costs $2000, and whilst technically even a single AA unit would be sufficient to bring one down given enough time, it would take far too long were the Kirov to make it to the enemy base; as such, most opponents I played would quickly spend significantly over $2000 on anti-air units to protect their base, giving me an immediate financial advantage (as well as the secret weapon of the Kirov). Even if I never did anything else with it, this forced investment on their part was often enough to make it a worthwhile purchase in almost any game I played simply due to a desire to immediately “counter” the Kirov threat.

A Kirov could therefore be used as a method to establish temporary financial superiority over an opponent who was forced into committing to significant amounts of otherwise-useless AA to counter the hypothetical threat. Additionally, this production of anti-air is likely to upset the production intentions of my opponent. There were many times when I found that an opponent was building up to a particular attack with a particular timing, but upon constructing a Kirov, it disrupted their original plan. This disruption seemed to be both in mechanic terms – they had to change what they were building – but also in mental/tactical terms, as it changed their expectations of the coming battle, made them rethink if they really wanted to attack immediately when there was an airship potentially bearing down on them, etc. Building a Kirov therefore forced them to commit both in-game money and real-world cognitive effort into the production of anti-aircraft weapons (and as we shall see the subsequent management of these weapons) and therefore, even if the Kirov was never used, its construction had significant positive effects by throwing a spanner into their plans, even if this was only to the extent of delaying them or encouraging a change in their build order. It was a small effort-investment in my part to force them to potentially rethink significant chunks of their strategy.

Destroy enemy AA


Once these AA units had been built, a strong secondary strategy was to set about destroying them. By destroying anti-air units in this way (units which are often weak to ground attack), I found I could force my enemy into building more anti-air units (which could then, once more, be destroyed!) simply to continue countering the potentially nonexistent threat of the looming Kirov that was somewhere on the map. It was a way to pressure my enemy into constantly spending money on anti-air units that might never see action and could easily be destroyed, rather on money to counter whatever other offensives I intended to launch. If they decided to stop spending on anti-air units and concentrate on the ground war – that, of course, was the moment to send in the Kirov. As part of the standard rock-paper-scissors game balance of many RTS games, AA units were generally highly vulnerable to ground attack and therefore open to attack, making this tactic all the more viable.

Force opponents to play differently

As well as forcing foes to “waste” money on the initial defence, and subsequently forcing them to continue wasting money on anti-air, the other significant impact of the Kirov was its use as an “abstract” weapon – something that may never be used, but scared my opponent for as long as the possibility existed. Some players would spend far more time scouting the map to prevent any easy paths of shroud for the Kirov to reach their base – although probably the optimal response, this still distracted them from other matters. Other players would also allow the threat of the Kirov to entirely change their gameplan. Many AA units in the game double as troop transports, and many players would commit to significant production of these units. If the Kirov appeared, they could defend, but if it didn’t, they had value as dual-use units. These units were however significantly weaker than “ground only” units, meaning that my opponents would reduce their offensive capability in order to account for the potential need for defence. A small number of players would consider the Kirov to be such a threat that they would overextend themselves and go out of their way to destroy it, potentially opening them up to easy counterattacks; an even smaller number would ignore the Kirov (considering it to be a joke unit), and they were naturally those who eventually got bombed by it.

In this way I would suggest that the production of a Kirov is akin to having a charged attack in a fighting game that is ready to be unleashed. The threat of the attack and the changes it forces into your opponent’s playstyle can be just as useful, if not more useful, than the attack itself. Although the Kirov was imagined by the developers as a purely offensive unit, it also had rare but surprising value in a defensive role due to the functioning of one of the game mechanics – how units automatically choose targets if no specific target is set. If attacked, a Kirov would easily soak up damage and distract enemy units from attacking your more important defensive units, and if not, the Kirov still survives to continue to attack. As with everything above, the balance for the Kirov remains between its threat and its use – getting this balance right and using this balance in a range of different ways unveiled the possibility for a surprisingly strong set of tactics with a seemingly worthless unit, and showed that units of this type can be used as much for their psychological as their mechanic value. If there was no use for the Kirov, keep it back as a threat; if the Kirov could safely attack, use it.

Concluding thoughts

SovThere were, however, limits to these ideas. Firstly, these tactics had limited applicability against an Allied player. The Allies possess a building which reveals the entire map to them, and thus any sly movements with the Kirov would be instantly revealed. However, a decent number of top players either played as the Soviets (mostly for their early-game rush abilities) or Yuri (for their seemingly-unstoppable mind control units and excellent late game options), and against such players the use of Kirovs in this way was supremely viable. Secondly, these tactics were better on some maps than others, and there were not many maps in the 1v1 ladder list (eight in total, if memory serves). Maps with oceans around them were ideal as few players took advantage of the naval game; and larger maps with no essential or strategically useful terrain at the edges were also useful for sneaking Kirovs past. Other maps, however, made it trickier to get them anywhere near my foe’s base before discovery. I therefore suspect that among the players who seriously competed for the highest ranks, some of those players who chose to play as the Allies may have never encountered any of the strategies I’ve described here, despite the many hundreds of games I played on the competitive ladder.


A perfect map for Kirov use; plenty of ocean and an apparent emphasis on land combat.

Despite the strength of these strategies, I don’t think I ever built more than one Kirov. The returns from the strategies here diminished rapidly the more you built – the threat of a single Kirov was sufficient for almost everything described here, and whilst the rare actual base attack with the Kirov would surely be stronger with two, the presence of two created rapidly diminishing returns. Although in friendly games I experimented with multiple Kirovs, a single Kirov remained the best for competitive play – it allowed me to psych out my opponents, to push them towards non-optimal builds and spending, and if all else failed, gave me a pretty powerful unit that once or twice actually saw action, rather than just floating above the battlefield as an existential threat that may one day descend from the heavens to wreak havoc. This set of interrelated strategies for this mobile semi-superweapon were something that, to the best of my knowledge, I pioneered. I never saw Kirovs in competitive use at a high level before this, and whilst once or twice I did see them used towards the end of my time playing the competitive ladder, I think I hold a strong claim to being the developer of these metagame ideas. Without the sound effect the Kirov would be a more effective unit in traditional battles and more able to sneak up on your opponent, but with the sound effect, it became a strikingly effective form of psychological warfare and opened up a wealth of interesting tactical options.

New House and New Markets

Just a quick one this week. First off, I’ve finally found a place to move to, which is this lovely building:


The intention is to move on the 3rd/4th of December, so we can therefore say my full-time development year will be starting on the 5th of December! More information nearer the time. I’ve also been planning my travels in this coming year and there are potentially a lot of places I’m going to be visiting to give talks (including this year’s roguelike conference, unless it ends up in North America), so more on those as and when – I’ve loved meeting fans in the past, so it would be awesome if anyone was based near some of the places I’ll be travelling to.

More immediately, I’ve redone market districts. The playtesting I’ve been doing has shown a serious issue in the navigation of market districts above all others – it was very unclear what the boundaries of the district were, it took a long time to find all the shops, there wasn’t much else to look at within the districts, and (quite simply) there were just too many damned shops. Here’s what they used to look like:

As you can see, there’s a huge number of shops (the normal buildings) and warehouses (the enclosed/walled buildings), but it’s not clear if you’d have missed one or not without exploring the entire district because they are so spread out. That means optimal gameplay within a market district is uncovering the entire thing rather than having some idea where to go for the shop you wanted, and that’s just not on. Instead, I decided to blend market districts with some of the housing districts, to create something like the version below. In this model (there’s naturally a large number of variations) the markets and warehouses that service them are placed in a pattern in the middle of the district. All the patterns make it very clear when you walk around them when you have “exhausted” the shops that district has, and also, this new contained version allows me to simply have fewer shops in a market district. They can now spawn with 7-10 shops (towns have only 1-3 shops, heavily biased towards 1 or 2), and always spawn with a currency exchange (the diamond building) and an auction house (the flattened cross building).

New_SThese both look a lot better, and more importantly, play a lot better. As above, there’s never uncertainty once you’ve seen every shop there is to see, and the smaller number will make the variations in shops between market districts much more significant, as in the past most districts contained most shops, which doesn’t really fit the scarcity gameplay mechanics one is used to from roguelikes. Alas, that’s everything for this week, but we’re only about a fortnight to the release of the colossal 0.6. There’s a couple more graphics I need to make, and then it’s just onto bug fixing. Stay tuned!

ProcJam and London Trip Report

Last night I returned from London having spoken at two conferences, one of which was the ProcJam introductory lecture series I’ve been posting quite a bit about in the last few weeks. Since I haven’t had time to get much coding work done this week I thought I’d post a trip report on this excursion instead!

The first of these was a conference on testing, practice and rehearsal. It is rare for a conference to be about a topic rather than about a particular part of a field, but this was one such example. The conference call was (deliberately, I assume) broad enough to include a wide range of scholars, ranging from me in the social sciences to a number of people from management, all the way to several people in the performing arts, theater, media, and literature. This led to a bunch of very interesting talks, but also many instances of that classic academic faux pas where two people from two fields are using the same word or term to mean two completely different things (this cropped up a lot with the word “rehearsal” in particular).

My talk was entitled “Predicting Play: The Expectations of Game Playtesting” (all academic papers must have a colon in them somewhere). The first half of my talk was an introduction to playtesting (which I felt was particularly appropriate given the range of academic fields present), but the meat of the talk was about how playtesting tries to assume and predict how users and players will interact with games and pieces of software, and most importantly, how these assumptions can be deliberately or accidentally undermined. To explain this I focused on three things – glitches/bugs, modding, and game guides – and how each of them challenges playtesting expectations. The talk went down very well and resulted in the potential for some very fruitful future opportunities and/or collaborations with others at the conference who were interested in my work (potentially involving a talk in Europe and an exhibition?), so more on this as and when.

After that was what PROCJAM. This was an awesome event organized by Michael Cook (who writes the Saturday Papers, one of the game blogs I read regularly) and, as part of the “opening” of the game jam, featured a total of seven talks (if I am remembering correctly). This included a talk from Gillian Smith about unusual ways to “do” procedural generation; a talk from Darren Grey (who I assume everyone who reads this blog knows) about player interaction and involvement in the proc-gen process; a talk from Tanya Short (sadly with Skype trouble) about semantic meaning in proc gen, which is strikingly close to a lot of things that really interest me; then a great talk from Hazel McKendrick about No Man’s Sky and “procedurality” vs “randomness” and different layers/scales of generation, which led very nicely onto my talk about actively designing a game around connecting these many layers.

I talked – as someone said on Twitter – at about “a mile a minute” (this is just how I speak!), but I thought the presentation went great (and I’m told Twitch chat responded very positively to some of my demonstrations, which is hugely gratifying). I believe we averaged around 150-200 viewers throughout the talks, which is an incredible turnout online for an event like this! This was in addition to around 50 people in person, so I’d say we peaked at around the 250 mark. After my talk I fielded some very interesting questions (and then some more off-stream from interested people after everyone else had gone for coffee). After my talk we then had a talk from Tom Coxon about generating lock and key systems, and a very, very cool presentation from Fernando Ramallo who sadly couldn’t be there in person, but sent an extremely interesting video in his stead. We then wrapped up and went to the pub for several hours, most of which I spent talking to Hazel, Gillian, and a range of fans who had attended and wanted to fire questions at me, and had a delicious pumpkin chilli rice thing. I then wrote this blog entry on the train back, having also been informed by my future housemate that we may have found a house to finally move to for my full-time development year! So that, also, is awesome. It’s always amazing to meet fans (even if they weren’t fans before the presentation!), which is why I’m going to continue posting any public speaking stuff I do here on the blog, as I’d love to meet anyone and everyone who follows this blog if you can find your way to somewhere I’m presentating something. There’s maybe as many as half a dozen conferences/game expos I hope to attend in the coming year in various points of the globe (at least one in US, one in Canada, maybe one in Hong Kong), and more info on those later.

All things considered it was an excellent trip, with ProcJam as an especially excellent highlight. That’s the most people I’ve presented to yet (though I aim to beat it!) and I’m so glad the response was so amazingly positive. You can view the video on Twitch here (my talk starts around 2:43, but all the talks are worth watching!), and I believe Youtube videos will be up very shortly (I’ll link them here once they are). You can download the playtesting presentation below, but my ProcJam one seems too large for WordPress to accept, alas. I never write myself notes for presentations because I believe in the power of improvisation and find that rote-learned talks often come off rather dull and stilted (at least when I do them), so there’s no commentary, but I’m sure you can figure out roughly the themes I was talking about. Hope you enjoy. This coming week is going to be spent doing all the admin stuff for this flat (and hoping our application isn’t rejected!), more final PhD edits and some URR 0.6 bugfixing. Next week: ANOTHER ENTRY ON SOMETHING!


The Value of a Good Door

0.6 is nearly finished. There’s very little left to do – I need to finish off the changes I want to make to market districts (shouldn’t take more than a day, or two at the outside) and then it’s onto bug fixes, glitches, edge cases, etc. Every city district except docks will be open for visiting this release – I’m leaving docks until we have NPCs and ships going around the world so that I can implement it all in one go. Equally, things like generating memorial statues and so on are going to be left for a little bit. For 0.6 there’s quite a lot of small bugs that do need fixing, however, so given the volume of remaining work required – and the fact that for a lot of this month I still need to finish off my doctorate and present at two conferences, one of which is the ProcJam in London – I think a release around the middle/end of November is realistic, with the 0.7 release after that aimed at a much smaller timescale (two months instead of seven, say?). A lot of these bugs are minor fixes, but a few – a particularly stubborn issue with road generation, some strangeness involving territorial expansion, towns not generating properly 100% of the time when they’re at the “end” of a road, etc – may take a little longer to divine the causes of.

Anyway, this week I’ve finished off the remaining procedural graphics for 0.6, which basically means doors. This took me about two days of graphical design. There’s now over two dozen different designs for doors when you look them up – the rest of this brief entry is four different screenshots. Two are from feudal cities, one from a polar hunter-gatherer settlement, and one from a graveyard. Some buildings share the same door patterns, though most are distinct and varied. I’ve also put some finishing touches to city centers, fixed a couple of the easier bugs and some typos, written up a new guidebook entry on “Buildings” for this release, and also done a lot of lore writing/planning which will begin to slowly seep into the game in the coming few releases. Additionally this week I’ve been submitting a vast number of abstracts to game studies conferences in the coming year, so that’s taken up a fair helping of my time too. Without further ado, here’s the four screenshots – hope you like ’em, and see you next week.





City Centres Part II: The Drastic Improvement

Y’know, looking at city centres now, I’m almost embarrassed by how plain and downright uninteresting they looked last week. Nevertheless, I suppose this serves as a good indication of how much can be done with a week of time, some thought, and some excellent suggestions from my fans (even if a lot of that week was spent house-hunting and a wide range of exciting vaccinations). So, here’s how centres currently look. Those that have a cathedral (a Vatican-esque building, only one exists for each religion, in the home civilization of that religion, and if the religion is a theocracy, then the civilization is ruled from there) look like this:


…whilst those without a cathedral look something like this (the core “wall”/structure is the same shape, you will note, as the corner towers and the gatehouses in each district, which is different for each civ):


There’s a range of buildings here, and each of them I tried to make look visually distinct in order to both add variety, but also add in identification when you may encounter a number of these potentially quite large structures exploring a city centre.


City centres contain a few embassies to other civilizations. These are the only buildings with walls around them – I reasoned other civs would want to have some sense of security for their footholds abroad. They also have a pair of flags outside (the white symbols) which will show you what nation that embassy belongs to. I also decided it would be interesting if they used the brick style/colour of that nation, not the nation of the city centre – as you can see in these pictures, this combination and the flags make embassies very easy to identify. At the same time, I had to make sure there were never too many in any one city, so you can’t learn about too many other civilizations at once. There will never be more than three embassies in a city centre, and two is around the average. They also have their own gates leading into the embassy grounds.


For courts, I found myself thinking about the shape of courts in the real world, and I came across the Star Chamber (mentioned in the Baroque Cycle, a secondary inspiration for the game). From these I decided to have courts roughly follow variations on the shape of a star, some more circular than others, some more pointed than others (though truly circular buildings are generally arenas in lower-class districts). In the first picture the court is the second building in the last column, and in the second picture it is the top-left-most building. These only appear in civs with certain justice policies, and they may be hubs for information like the wanted level of certain NPCs, bounties, and might play a role in any future legal systems that the player can fall foul of.


I’ve added guilds. Currently these are mercenary guilds, from which you’ll be able to acquire the most expensive and best allies (vs taverns, slave markets, etc). Fancy recruiting from the Legion of the Black Flame or the Chapter of the Bloody Fist? For these I drew inspiration from real-world guild halls which often consist of several buildings over several layers (there are many where I live in York) – guild halls are thus buildings with many “layers”, and you can see one in the top-left of the first picture, and the second across on the top row in the second picture.

Slave Markets

These will crop up (as you might expect) in slaving civs – they will be closer to the open-air markets in nomadic fortresses than the enclosed shops that predominate in feudal nations. There aren’t any visible here, but they consist of a small number of thick, straight lines, with clear “market areas” at their intersections where you’ll find vendors in 0.8. Where guilds offer skilled and well-trained allies, slave markets will generally offer somewhat less competent allies, but cheap… though that’s not to say some skilled allies might not have been trapped into slavery here and there.


Art galleries. These are generally either an L shape (start of third row in picture 2) or a U shape (third down in third column in picture 1). These are similar-ish shapes to stables, but you’ll never find a stable and a gallery in the same district (and galleries, like most other buildings, have signs outside to denote their function). They are going to contain paintings. These are going to be awesome.


Several people suggested these, and it fits in very well with the future history-changing mechanics. These were inspired some of the larger real-world memorials which are more like something you walk around, though these also have a “statue” in the middle which will be related to a historical event. In the top picture you can see one third in the top row, and in the second picture one is third on the bottom row. Statue generation will not be fully present in this version, in the interests of actually getting it released before the end of November.


All city centres have a mint for the bank of that nation – these are similar to banks in that they are built around “blocks” and a square-zigzag pattern. The bottom-right in the first picture and the bottom-right in the second picture give some examples.


NewSFor those nations with a democratic preference, they are ruled from Parliaments. They’re based primarily on the UK Parliament building, and have “corrugated” walls, and often clock/bell towers and multiple entrances. Here’s an example from another city on the left. Like many buildings in city centres I would think Parliaments will have some guards patrolling outside them, and might have vaults underneath containing something. This is around a fifth of all civilizations, so parliament buildings are relatively rare buildings to crop up in city centres. Parliaments, like Mints, take up two “blocks” of a city centre, whereas all the other buildings listed here only take up a single block. There are also gardens and lakes around most city centres too, though those particularly packed with buildings may have little room for the greener things in life.

Wonders of the World

There will also be fifteen super-special buildings, approximately one in each city. In a few worlds one or two might not generate, or one or two cities might be without a wonder. These will be buildings that are sometimes in the centre, sometimes elsewhere, which are special and unique (the Panopticon in a previous post) and are to do with the story. More on these as and when.


City centres are now actually interesting to walk around. I need to work on flag generation, but that should be a pretty snappy task. Otherwise they’re pretty much finished. Hopefully next week (or at worst, the week after) I should be able to unveil an image of a complete city, which should look bloody amazing.

A final note. Development is now about… 80% of my activity? I’m not really full-timing it yet, but I’m not far off. There isn’t much more doctorate that needs doing, most of my time is spent waiting for feedback and then sending in more edits. Moving house is, as things always do, proving trickier than expected, but we’re still hoping to move in November. Lincoln (where we’re moving to) seems to be a town inundated with flats with low ceilings, and for someone of my elevation, that is sadly not workable, not to mention that almost every flat seems to come with damned tiny bedrooms. Development is going to stay rapid until then, though, so don’t think the fact I’m not full-timing yet doesn’t mean we aren’t back to weekly URRpdates (I remain oddly proud of that term), because we are!

(Lastly, yes, I do know I switch between the UK/US spellings of centre/center constantly in these entries. FORGIVE ME)