Thursday, October 24, 2013

Game is harder: Part one

Luna is 12-3 in competitive Dota 2 since the patch was added to the client - which is very impressive when contrasted against a 45% total win rate [including these 15 results]. Everyone is trying to make Nightstalker and Lycanthrope work, Shadow Shaman is beginning to re-emerge as a competitive pick, Pugna is actually a thing, dual lanes are starting to become popular, and Lich has gone from a non-consideration to an autoban or first pick.

Big patches are always exciting for me. Everyone is talking about the new patch and trying to analyze the extent to which the game will change and in what ways it is about to change [or has already changed!]. I like to think that metagame in Dota is sufficiently complex to make it very difficult to predict changes purely by looking at changelogs - Icefrog and co., by virtue of the way patches are released, seem to agree. We are likely to see small tweaks to various of the latest changes in the upcoming weeks. Which is to say, a lot of the time, working out how the new metagame is going to look is more a matter of playing it out than a matter of thinking it through. There is therefore a very real limit to the amount of meaningful analysis one can do on this sort of topic. That said, I intend to give it a go here. However, instead of just pointing to a list of interesting facts about the patch, I plan to present a singular thread of conceptual change - particularly, I will argue that this patch is yet another step towards making the game less passive. In order to do so, I will draw on a multitude of changes from the changelog. Finally, I will defend the claim that this is a positive move for the game.

Before 6.79

First, I should explain very briefly why I use the qualifier 'yet another' above. I won't spend too much time on this but here are a few examples of how previous patches have tried to move the game away from more passive playstyles or mindsets.  To begin with, we've seen some controversial changes to Captain's Mode this year:

- Captain's Mode bans changed from 2/3 to 2/2/1
- Captain's Mode picks changed from 3/2 to 2/2/1

The change to CM which came before these had changed the beginning of the draft from 3 bans into 2 pick to 2 bans into 3 picks. This appeared to be a clear attempt to force teams out of their comfort zones, introduce more variety into the metagame, and guarantee that teams would have to face at least some heroes they didn't want to. The above change follows a similar conceptual motivation, I think. Here what the change does is make the draft more interactive. Because the bans happen in three stages instead of two these days, it's harder for a team to sit back at any point in the draft and assume they'll get the rest of the picks they want. There are three opportunities for the other team to interfere with your draft, two of which come with plenty of information about it.

Meanwhile, the June 4th changelog also made it easier to reach level 11 quickly:

- Level 11 XP requirements reduced from 6500 to 6000
- Level 12 XP requirements increased from 7700 to 8200

This change is almost irrelevant for carry heroes which don't usually rely on their ultimates. Moreover, carry heroes are usually happy to spend a lot of time passively farming - thus the exact time when they reach any given level is a lot less significant than it is for other heroes. So this change buffs solos and supports. And what do you know, these are usually the ganking heroes. Further, we saw a change that would anticipate one of the 6.79 changes:

- Deny XP and Bonus XP/Gold AOE is now the same as regular XP AOE (1000->1200)

The direct result of this being an easier time in the offlane for offlaners, who were struggling to avoid being zoned out of lane by creep pulls and harassment - thus leading to a strange situation in which teams more or less stopped picking heroes to actually lane the offlane. Rather, we spent most of this year seeing heroes which can jungle or interfere with creep movements going to the offlane. Making it easier to offlane increases the diversity of heroes that can lane there and increases the pressure on heroes farming in the safe lane, thus reducing a team's ability to win a game solely via passive and controlled farm.

In addition to these changes, many others have been made during the course of the year in a similar vein - most notable are those which rearranged the structures of gaining gold via ganks and retaining gold [in terms of reliable and unreliable gold]. There seems to be a resounding sentiment in Icefrog's team that the game is better where it's less passive.


Here I will discuss 10 changes from the 6.79 changelog, grouped under two broad headings of 'Map changes' and 'Item changes'. Specific hero changes will not be discussed during this post - though I would suggest that even those are in support of my thesis statement here.

Map changes:

1. Creep pulls are less relevant early on. This is evidenced by two changes in the changelog.

* Swapped the medium pullable Neutral Camp with the small camp

This change means that supports will struggle to deny exp to the offlaner and get exp for themselves quickly, even if they stack the camp before pulling. But this is not entirely a nerf to supports who will be able to clear these camps faster if they want a quick level 2 before ganking.

* Non-Ancient Neutrals now split XP with all heroes in the AoE instead of just the team that killed them

This change means that offlaners who manage to not get completely zoned out can gain exp just by being near enemy creep pulls and can thus get denied significantly less in the process, undermining one major purpose of the pulls. Perhaps more significantly, as Akke emphasizes, this change makes it a lot easier for teams to use aggressive tri-lanes. Where supports might be able to zone out an offlaner, they won't be able to zone out an enemy trilane - meaning that 3v3 lanes will not be decided by safe creep pulling but rather actual interactions between the heroes.

2. Offlaning is now a lot easier. This is directly related to the changes above but also supported by two further changes in the changelog.

* Creeps now meet a bit closer to the tower in your offlane

The changelog here is perhaps understating this change when it says 'a bit'. The change, in fact, is significant enough such that pro teams are beginning to consider blocking creeps in the safe lane, despite all the potential issues that come with that. Either way, it's a lot harder for the safe lane farmer to simply hold creep equilibrium at the most beneficial position in lane without giving it a moment's thought. This gives offlaners more leg room to get exp early on [and also improves aggressive tri-lanes or dual lanes for obviously reasons].

* Ranged Heroes now get the same denied experience as melee heroes (instead of less) 
* XP AoE increased from 1200 to 1300

These changes echo the sentiments of the similarly flavoured change from the June 4th patch mentioned above. However, the previous change was not extreme enough. Despite it being easier in theory to lane more things in the offlane after the June 4 change, in practice the offlane continued to be a lane which wasn't actually laned in but rather served as the lane for which you'd need a unique plan for in any given game. Increasing the AoE for gaining XP complements the above changes very well and makes it near impossible for an offlaner to be completely held to level 1 with zero experience at the start of the game. Meanwhile, the more interesting change is the one which levels ranged and melee heroes in terms of XP denied. While the status quo before this change was a huge deterrent against picking ranged offlaners, with that deterrent gone, ranged heroes come back into play there - even if they can't perform special tricks like farming ancients or jungle. One upshot of this is that carries will be finding it a bit harder to farm uninterrupted in such situations, unless their supports are permanently at hand.

3. Buyback is nerfed as a tool for passive control. While buying back is still an option, it now comes with a very serious cost. Icefrog has repeatedly tried to nerf buyback and failed to achieve the desired result but this time I think he's done a great job.

* Buyback prevents gaining unreliable gold (creeps, neutrals, etc) until your normal respawn time finishes
* When buying back, 25% of the remaining respawn time will be added to your next death

Both the changes above result in a situation where it is only profitable to buy back either to rejoin an ongoing fight, or out of necessity to defend or attack some key objective. We certainly will not be seeing people buying back willy-nilly [and rage-buyouts in pubs will be significantly disincentivized]. This change is excellent in that it manages to preserve buybacks just as far as we value having them in the game. If a team plans carefully with their gold and takes particular fights on the basis of having buyback gold [cough, Alliance, cough], surely they deserve some reward? So buyback is not nerfed enough to make the clutch buybacks less clutch but nerfed just enough to make any other kind of buyback into a very big decision. The direct consequence of this is that teams have one less way to passively control the game just by having more gold.

4. Roshan's respawn time is unreliable. There isn't an awful lot to say about this one. As above, reducing the certainty a team can have of what time Roshan will respawn reduces their capacity to passive transform advantage into further advantage. Guaranteed Roshan kills is just another way that a team could grind out advantages until their economic advantage was eventually too significant to be able to defend against. 

* Roshan will respawn at a random time between 8 and 11 minutes after death

Of course, both teams will still be writing down when Roshan dies but now they will have to act as if Roshan is alive from the 8 minute marker, despite his only respawning any time within the next 3 minutes. This may force a team that really needs Roshan to commit a lot more time and resources to it and thus offer the enemy team some breathing room to catch up. Meanwhile, it also increases the chance of a team winning a fight and then getting Roshan, without the Roshan being an intended consequence of the fight itself. While some may think this 'randomness' is undesirable, I believe that some degree of randomness is necessary for the game to remain complex and exciting and I think randomness in cases which usually favour the underdog is the best kind of randomness to have. And yet, Icefrog thought of this too, to counterbalance this one somewhat:

* When Aegis expires unused, it heals the hero fully over 5 seconds (regen dispels on damage from players or buildings)

So while it's harder to force Roshan kills, the reward of killing Roshan is greater. Enemy teams that correctly anticipate the time that Aegis runs out and pounce on the Aegis holder as it goes down will still be able to kill that hero but in all other cases where Aegis is unused throughout its duration, it can still offer a reward to the Aegis holder. This is especially significant in cases where the Aegis holder makes a good decision [and certainly the kind of decision which encourages active engagement rather than passive control] to play aggressively in order to make use of the Aegis but perhaps just doesn't quite manage to die.

5. Gold acquisition is faster. Heroes getting gold slightly faster will probably not make a huge impact on core heroes. However, supports benefit massively from this.

* Gold for ending a spree changed from 75->600 (3x->10x) to 125->1000* Your gold income is now 1 per 0.6 seconds, up from 1 per 0.8 seconds

The income change means that all heroes will have gained 1000 income gold after 10 minutes. Previously this figure would have been 750 gold. While 250 gold at the 10 minute mark might not be significant for core heroes, it's extremely significant for supports, as is the difference at the 5 minute mark which is 125 gold. This means, for example, that support heroes which would usually get into a position where they might have to decide between boots and more wards at 5 minutes will now be able to buy both. Meanwhile, supports will continue to benefit from this income throughout the game, often as their main source of income and thus the effect of this change cannot be understated. Supports being less poor means supports being less useless which in turn means supports being able to actively do more things in the game throughout the game. Lastly, the increase to how much gold you get for ending sprees echoes sentiments above in that it increases the importance of decision-making throughout the game, rather than weighting decisions made earlier above those made later. It's now a bigger deal if a core hero who has enjoyed major success all game slips up at any given point in time and dies. While it's reasonable for a player's early good decisions to act as insurance against their later bad decisions, the situation needs to be more balanced than it's previously been and these kinds of changes do well to work in that direction.

Part two can be found below or @ In part two I discuss the effect of various item changes in 6.79 and conclude by arguing in defence of changing Dota into a less passive game.

Game is harder: Part two

Item changes

6. Wards are cheaper. No they're not, you're probably thinking. Well yes, they aren't directly cheaper, but they will work out to be cheaper given that they last longer and given that you will now sometimes get returns on your [sentry] wards.

Observer Ward
- Duration increased from 6 to 7 minutes
- Killing Observer Wards now grants a 50 gold bounty
Sentry Ward

- Duration increased from 3 to 4 minutes
- No longer gives permanent vision around it
- Gives ground vision in a 150 area around it for 12 seconds when placed

So lasting longer puts less pressure on supports in terms of gold but where it may seem like this might make Observer wards do more to prevent gangs from being successful, Sentry Wards save the day. This is because discovering enemy Observer Wards now rewards you with some of your money back. That said, Sentry Wards are also nerfed in that they only grant vision temporarily after placed. This is also an expertly designed change as they still provide enough vision to be used on their own to kill an observer ward they were placed sufficiently near to. However, they no longer have the strange property of sometimes being able to scout an enemy gank on the way in - a property I call strange because it's totally separate from their purpose. 

7. Smoke of Deceit is buffed. Related to the above but taken even further, Smoke of Deceit is now an even stronger ganking tool.

Smoke Of Deceit
- Smoked units are now always hidden until the buff is removed (rather than being hidden only from minimap and partially from world)

This means that judging when another team is Smoke ganking is now purely a matter of skill. The only way the game will directly tell you that they are Smoked is at the point at which it's too late and the Smoke has already broken. Again, this reduces the extent to which a team with a gold advantage can passively control the game as nobody is ever completely safe while enemies are missing from the map. 

8. Urn of Shadows improved as a ganking tool. While Urn of Shadows has never stopped being relevant, it has lately become more of a healing item than a killing item.

Urn of Shadows- Soul Release cooldown reduced from 10 to 7

By reducing the cooldown of Soul Release, Urn becomes a much more aggressive item, extremely well suited to regular early game ganks or skirmishes. 

9. Bottle-couriering is even further nerfed. Again, this is something that was attempted in the previous big patch but seems to have more or less failed. At the time, I was hoping that the loss of speed on couriers when carrying empty bottles would result in mid lanes becoming more interesting again. It turns out the nerf was not enough and Icefrog is back for more here. 

- Gold cost increased from 600 to 650
Flying Courier

- Cannot be purchased until 3 minutes after creeps spawn

The slight increase in the cost of bottle together with the fact that couriers have to walk for 3 minutes should hopefully do significant work to reduce the status of mid lanes as being largely defined by bottle-couriering. Meanwhile, supports now have a much bigger incentive to try snipe courier in the first 3 minutes, especially in cases where the enemy mid laner is unlikely to have a good time without sending in items before the 3 minute mark. This not only creates a more interesting mid lane - one which can actually be won or lost - but it gives supports yet another active thing to do in the early game.

10. Regen items for carries are nerfed. This comes in two forms. Firstly, the main boots of choice for carry heroes, Tranquil Boots, are no longer likely to be popular for carry heroes, given the fact that they will almost always be broken in lane and thus won't provide any regen at all. That said, I did see Burning going Tranquils on Luna immediately after the patch and there definitely is some argument to be made for why it's still a good choice - being cheap and efficient and still providing plenty of movespeed even when broken.

New Tranquil Boots
Active Boots:
+ 85 Movement speed
+ 4 Armor
+ 10 HP Regeneration

Broken Boots:
+ 60 Movement speed
+ 4 Armor
Restores when you haven't attacked or been attacked in the last 13 seconds.
Cannot be disassembled

The fact that Tranquil Boots can no longer be disassembled is again a nerf targeted at core heroes who had intended to use some of the components later for a Force Staff or a VladsSupports, on the other hand, will be more than happy to keep Tranquils as their boots for the entire game usually. And of course they will thrilled to have such insanely high move speed and will not need to attack nearly as much as cores and thus be able to make use of the regen very well.

Lastly, Tangos and Salves [and clarities] have also been changed to make it harder to pool heroes with early consumables. While it is not impossible to pool cores with Tangos - and the scene has adapted to do this very quickly - it is impossible to pool Salves. This might not always be relevant, but given that Tranquils are less desirable a choice for carries now, not being able to have extra salves seems to drill the point home - that staying in lane after almost dying or incurring serious harassment is significantly harder if you're on your own.

So why is a less passive Dota a better Dota?

In reality my major aim in these two posts has not be evaluative - rather, I've aimed to make a case for interpreting the recent changelog as representative of a certain broad agenda for the game among Icefrog and co. That said, I happen to support this agenda myself and I did promise to defend my position that making Dota less passive is better for the game and will thus briefly outline an argument for this position and consider a counter-argument to it, explaining where I think the counter-argument goes wrong.

My argument here is really quite simple. I think that a game which is more competitive is equivalent to a more competitive game. While this might sound like a tautology, it isn't one. The former refers to instances of the game having high levels of competition in them while the latter refers to the game as a whole being more competitive. In other words, I think that the Dota 2 industry is a more competitive industry if more games involve fierce competition. I take for granted that it's a good thing for Dota 2 to be more competitive over all, given the nature of the industry. Thus, unless there is a case to be made against my initial claim here, it follows naturally that what we want is for the game to be such that it makes each instance of itself more competitive.

This, I believe, is least the case when the game enables or the metagame encourages passive play. A lot of what people have come to call 'Rat Dota' is related to this - though I do not have a particular objection to a style of play which involves, say, split-pushing. That said, for the longest time, Dota 2 has moved in a direction which rewards teams for taking early advantages and then consolidating those advantages, clinically and incrementally gaining more advantages until finally they have so much advantage so as to essentially 'automatically' win the game. The problem I have with this situation is that it weights a lot of decisions as far more important than other decisions when, on the face of it, they shouldn't be. Of course it is sensible in any strategy game that making good decisions early on puts one in a better position later on - but I dislike the idea that making good decisions early on could put one in such an advantageous position that the other team can make significantly better decisions than you later on and still lose the game. I think that ideally we want the team that wins a game to be making better decisions than their opponents throughout [at least most of] the game and not just at the beginning of the game. And this is exactly what the 6.79 changes encourage. Supports now have to actually do things to be relevant early on instead of being able to default to safely zoning enemies and pulling creeps to slowly gain passive advantage. Killing Roshan each time he spawns is harder to count on, players are more rewarded for making big kills on heroes which are dominant and it's harder to rely on securities such as buyback to use your finances in a direct way to passively stay ahead.

A counter-argument that I've already encountered on this subject suggests that many of the things I'm referring to as 'making the game less passive' are, in fact, just making the game 'more random'. But surely randomness is bad in a strategy game? Isn't the entire premise of strategy games that good decisions be what is rewarded and not luck of the draw?

Absolutely, too much randomness is a problem for a strategy game - there is simply no denying it. However, a more nuanced approach to this particular case would recognize that the 6.79 patch has actually reduced all sorts of bad randomness - look, for example, at all the changes to neutral camps. Now you are more or less guaranteed to get the creep you need in a short space of time on Chen, Enchantress, or Doom. Meanwhile, farming the jungle has become a lot more consistent in terms of how hard each camp is to kill and how much each camp rewards you. This is an excellent change as we don't want the random fact of which camps spawn at which times to have too much effect on a game. So which things which I praise are actually being accused of being random [presumably in a bad way]? To give some examples, one might suggest that the fact that couriers can now get sniped more easily or the fact that Roshan's spawn time is uncertain means a net increase in randomness in the game. The problem with the latter complaint is that Roshan's spawn time is not completely random but random within a known window and thus players still have more than enough control to make good decisions about how to deal with it - these decisions need only be more complex now. This is a good thing. Meanwhile, the former complaint just gives players no credit at all. Sure, couriers are more susceptible to being sniped now. But whether or not a support succeeds in doing this is not a matter of chance - there is a calculated decision involved in deciding how long to linger waiting for a courier and whether or not it's worth the risk. This is no different to any decision made by supports to do any other kind of rotation for any other kind of purpose during the game. 

One final point I should briefly address which was alluded to in part 1 and has now been alluded to again is the notion of randomness which is not bad. Here I simply bite the bullet. Yes, the strategic element of the game is probably going to be somewhat undermined by any and all forms of randomness. But there's a reason we play Dota and not Chess. And as someone who grew up playing Chess, I honestly think it would be a lot more interesting and exciting game if it did allow for some small degree of chance. Look at most successful strategy games - computer games, card games, board games - and you will see very easily that they tend to contain at least some degree of chance. This is no mistake. While it might reduce the strategic control somewhat, the control you do have becomes a lot more complex and interesting. And it sure as hell makes for more fun.

As always, thanks for reading - and any feedback is greatly appreciated!

Wednesday, October 9, 2013

A Defence of Regular Roster Changes

Update: Some of the images I tried to hotlink have been giving me trouble so I've added superior versions of each of them in order to preserve your viewing pleasure. Sorry, I'm really bad at formatting stuff. Still quite new to blogging.

There is a very long-standing sentiment in the Esports community and especially, I think, in the Dota community that regular roster changes are a bad thing. By 'regular', I'm talking about 1-3 times a year. Obviously teams changing their rosters on a weekly basis is problematic, so I won't be defending that at all here. Why do people not like it when teams make roster changes? Well, it's unprofessional, we are told. It appears to show instability in the given team and, accordingly, in the given scene. Lastly, perhaps the strongest sentiment around this issue is that it shows a lack of commitment of players to their teams. Surely these teams have not worked hard enough at being a team if they want to make changes to their roster after only being together a few months - they are just taking the easy way out and this is deplorable. These complaints are clearly all related and I think they are all generally misguided. By the end of this post I hope to have presented a strong case for why regular roster changes are a good thing in the professional Esports scene.

1. Regular roster changes are a sign of professionalism, not a lack thereof

One need look only as far as Professional Sports to realize that player transfers are a necessary part of a properly functioning professional scene. If you look at the English Premier League, for example, you'd be hard pressed to find a season where any team doesn't add or remove at least a few players.

The fact is that players moving between teams is an indication of a competitive focus as opposed to a casual one. Why do people leave their teams for other teams? Well, because they think they'll have a better chance of winning, or more opportunity to improve on their game. Of course there are cases of individuals who are fickle or uncommitted and make rash decisions about switching between teams but it's hard to find any kind of evidence supporting this being the general case. Meanwhile, the only reasons I can see as to why a team might maintain the same line-up indefinitely are either that their focus simply isn't competitive [read: friendship] or that their focus is competitive but they just keep winning. The reality is that the latter case is historically unprecedented in both sports and esports. Even dominant teams often have to make changes in order to maintain dominance - in fact later I will argue that dominant teams have special reasons for making changes precisely because they are dominant.

If a Dota team truly has a competitive focus, they will have to ask themselves at least every few months whether or not they might benefit from making changes to their roster. In reality, the question is hardly ever whether or not to make any changes but rather how many changes are needed, how soon they are needed, and which players should be replaced or added. One concern resulting from this kind of reasoning might be that players seem quite unprotected in an industry where finding a stable income is very challenging. This is absolutely true and definitely a serious concern but it's one which should be addressed by sponsors and organizations. The Chinese are way ahead in this regard when it comes to Dota 2. Again, if we look at the professional sports world, we see that the way players protect themselves is by signing extended contracts. This way they are financially secure even if they stop being asked to play or their team decides to go another direction in their development. Of course, in order for this to be possible, the industry does need to reach a point where organizations can sustainably sign these kinds of contracts and what this means is that sponsors and teams need to start finding more ways to make a safe return on their investments. Again, as the Chinese have already begun to do, one way to ensure this is to actually enforce the contracts that do exist. When player transfers do occur, if a player being transferred is under contract, the team they are joining must pay a fee for their transfer.

All of this leads nicely to the issue of 'stability' and how players who move teams a lot are perceived as making the scene unstable. Again, here, I believe that once the scene is fully established, the importance of stability will be mostly in the main teams/organizations being stable. No sport gets ruined if some player in some team decides to throw his or her toys out of the cot. Provided the important institutions are stable, the industry can survive problems caused by individual players largely unscathed. What's more, the type of contracts described above include these risks. If an organization signs a contract with a player, they are agreeing to take on various risks involving the conduct and performance of that said player in return for the player agreeing to various obligations. So, a lot of what I'm saying here does depend on our industry growing beyond where it is and developing infrastructure to support enforceable contracts for all professional teams and a set of regulations governing the administration of players. That said, even though the industry might not be quite there yet, it is a mistake to think that this means it all becomes the obligation of the players. As discussed above, there are always good reasons for truly competitive players to want to change to another team and if an under-developed scene restricts the ability of players to focus just on competing, then this is a fact we should acknowledge but regret, not one we should take for granted.

2. Why the best teams might have special need for roster changes

A lot of the above analysis can be summarized into the basic notion that competitive players who know their team isn't the best team should probably do whatever it takes to make their team the best team. One of the things they can do is change the players in their team and thus we should expect roster changes in a competitive environment, rather than bemoaning them. What might be less obvious is why players who know their teams are the best teams should consider roster changes. Here I have arguments both for the general case and for Dota 2 in particular.

One of the main ways that anybody who works in a competitive environment improves is by competing against their superiors. It's a lot harder to learn how to get better without constantly exposing yourself to something that is better. This presents a problem for those who are already the best - because, being the best, their is necessarily nobody who is better from whom to learn. Of course it's not always that cut and dry and often there are many teams which are close enough to one another's level such that they can all learn valuable things from playing against and studying one another. Further, it is obviously possible to learn things even from teams that are significantly worse than one's own and also possible to learn valuable lessons purely in the abstract upon reflection or by work. It's just a lot harder and a much slower process sifting through mountains of sand to try find hidden gems. That said, in an environment where everyone is continually working hard to improve, it is a serious disadvantage to the top teams that one of the main methods of improvement is limited for them.

Moreover, whenever any team is established as the best team in a given environment, that team becomes the de facto 'team to beat'. What this means is that most other teams are spending a disproportionate amount of time studying the best team and working out ways to beat them. This is one of the reasons that a lot of players prefer to be underdogs going into tournaments. When you're the favourites, everybody knows you're the favourites. The spotlight is on you and instead of just having to work out ways to win, you have to work out ways to win against people who are prioritizing beating you in particular over winning in general. So, because the best teams are at a significant disadvantage in at least these two ways when it comes to maintaining their status as best, it should be completely unsurprising when a top team decides to reinforce by adding a new player or two. This both strengthens their team and blocks potential competitors from doing so.

Focusing on Dota 2 in particular, I think there are even more reasons that roster changes eventually become a necessity for the top teams. Here, again, I will present two such reasons. First, Dota 2 has a very healthy metagame. By 'healthy' here I mean that it changes a lot. This is partially because a lot of Dota's metagame is what I've previously described as soft metagame - and is thus subject to severe unpredictability - and partially because even those parts of it which are hard metagame, the parts dictated by the game itself, are changing fairly regularly. Because the game changes and because people try new things, the way people excel at the game is also constantly changing. How this relates to roster changes is in the sense that a player could very easily excel more in one metagame than another. Where this is the case, it is not strange to expect that player to be replaced after a series of metagame changes occur.

Secondly, one of the things that is so awesome about Dota 2 is its complexity. People have spent years trying to find ways to pull meaningful statistics out of the game - things which could be predictively relevant - and have largely failed to do so. Don't get me wrong, I'm a huge fan of quantitative analysis [I just do less of it because I'm not very qualified at it]. The reality, however, is that we're still far from even knowing which things are the important things to measure, never mind how to measure them. Why does the complexity of the game affect the need for roster changes? Well, one of the consequences of a game being so complex is that it is never really clear how close the best actual teams are to what the best possible teams might look like.

To illustrate, consider Lionel Messi - presumably the best Football player in the world at the moment. If one were to analyze Messi in order to find out just how good he is and how much he can still improve and in what ways he could improve, it probably wouldn't be that hard. We might say he's nearly perfect, he could score 100% of the chances he gets in front of goal instead of 90% and he might get to that point by practising more shots at goal in training or spending more time studying opposing teams' defences. Now obviously this is over-simplified. But the fact remains that we've produced an at least plausible over-simplification. For Dota, it's hard to know where to even start. How good is s4 relative to the best he could possibly be? How does one even answer that question? Dota players are constantly finding new ways to be good at Dota and thus we don't even know what a perfect Dota player would look like at any given point in time. Again, returning to the topic at hand, what this means is that there is always room for justification in changing players in a team, even if it's the best team around. Sometimes a team will just know that they've stopped being able to improve in one particular way they want to improve in and will have to acknowledge that the only way to proceed is by swapping someone out.

3. Case Study: TI3 and post-TI3 reshuffles

Alliance: Loda, Akke, s4, EGM, AdmiralBulldog
Following from the above, I thought it might be appropriate to give some examples of my arguments in practice. The top 3 teams at TI3 had a minimum of 2 new players acquired since TI2. Alliance was a completely new team, Na'Vi had added Kuroky and Funn1k, and Orange had added Kyxy, Ohaiyo and Net. Meanwhile, while Chuan bragged in an early TI3 interview that iG and LGD were the only two teams not to change their rosters since TI2, iG ended in tied 5th and LGD didn't even make top 8. Granted, both of these teams could have done better at the competition and both lost out to some bad luck and some big plays, not only to their own deficiencies. That said, it's also easy enough to understand why their TI3 was worse than their TI2 precisely in terms of their lack of roster changes. This leads me to what will be a brief analysis of some of the Chinese teams post-reshuffle and why I think they are better off now than before.

It's no secret that iG were having internal problems for most of 2013. Their results began to decline near the beginning of the Super League and never really quite recovered. But over and above results, the lack of team unity was clear in any tournament footage of them over the year, as well as various interviews with and posts made by Chuan. I remember observing a few months before TI3 [when I had planned to do a write-up about the fall of iG] that since TI2 they had won every series where they took the first game and lost every series where they lost the first game. To me, this is a clear indication of a team running purely on individual ability + confidence. The players aren't happy in the team they're in but as long as they keep winning, they're happy enough to . . . keep winning. The problem comes when they don't win things and this leads to them not winning even more things - and so on. In a team with such experienced players, it's simply difficult to accept that every loss would tilt them so hard if they weren't already unhappy within their team. Anyway, as we all know, iG changed 2 of their players after TI3 and since then have lost only one match vs DK, who I'm sure we can all forgive any team for losing against right now.

DARK SYLAR, formerly of LGD
Meanwhile, LGD is a totally different story. Here, the issue doesn't seem to have been personal. Rather, LGD, who have for a long time been seen as the quintessential Chinese style Dota team, spent much of 2013 trying to remove themselves from that stereotype, recognizing that their competition were improving quickly and that in order to stay near the top, they would have to learn to adapt and play aggressive style Dota too. After losing out against Alliance in the G-1 League Grand Finals, LGD moved things up a gear and were able to take 4 games [over 2 matches] off of Na'Vi during the Alienware Cup, despite ultimately losing out in the finals. More and more we began to see variety in LGD's picks and more and more we began to see them dabbling in aggressive picks. However, a by-product of this was that the backbone of LGD's old gameplan began to become less and less of an asset and more and more of a cost. I think this is really tragic, to be honest, because it's clear to me that Sylar was LGD's best player for a very very long time and perhaps even their best player at the time he left the team. In fact, if not for Akke's amazing Blinding Light vs LGD at TI3, Sylar might be an LGD hero right now instead of an ex-LGD player. That said, being the best player isn't worth anything if you can't play the way your team wants to play and it appears that this is exactly what happened with LGD. With the addition of Xiaotuji, LGD have immediately begun playing much much more aggressively and though their results have been mixed so far, it is very clear to any observers that they are at least one team with one plan right now, determined to develop in a very particular new way. Even if things don't work out immediately, this is a far healthier setup for a competitive team.

DARK DK: Dai, Burning, iceiceice, lanm, Mushi
Perhaps a brief mention of DK is also called for at this point. What did roster changes do for DK? Well we're currently experiencing the most incredible all-star team of all time - an enterprise which people are often skeptical about, but one which so far seems to be working really well for DK. More than that, though, Burning didn't retire from Dota 2. And you know what, if the choice was between staying in the same DK team or retiring, the evidence suggests that he probably would have retired. While this comes from a completely different angle, it is hopefully not controversial that keeping B-God active in Dota 2 is great for the industry.

Now you might think I've been quite selective here. Well, I have. You might think it's inappropriate because there are also teams who ended up worse off from the Chinese reshuffle. Well, to that I say firstly that roster changes being a good thing in the interest of competitive play does not require everyone to be able to end up better after a change. All I need to show is that it makes sense for each team to want to get better and one way for them to do so is by changing their rosters. It's absurd, in fact, to expect every one of them to be able to become the best at the same time. That said, I believe that every Chinese team, with the exception of perhaps [my beloved] Rattlesnake, has actually improved in one way or another after the reshuffle. I won't defend this position here though as this post is already ridiculously long. Relatedly, it's not clear that player transfers is a zero-sum game. It is possible for teams to exchange one player for another and both end up better for it.

Concluding remarks

I suspect that many readers will think that many of my points are not new or surprising. I actually hope this is the case. That said, while the sentiment that roster changes are a healthy part of a growing environment does seem to be on the rise, it is one which is not voiced often enough and one which I think is very important to acknowledge. Over and above all the arguments I've made in favour of roster changes in this post, I also just think people should be allowed to change their minds.

Tuesday, October 1, 2013

Coming Soon...

Anyone who has checked this page recently will have been disappointed to find no new content. This will change very soon. Starting next week I plan to post very regularly - at least for the rest of the year. So watch this space!

I am but a one man light brigade, yet I shall give my all!