The Evolution of Dota's Most Subtle Economy
As we saw quite a long break in competitive Dota since The International, I started thinking about the looming 7.23 patch, and the direction that Dota will take when it finally hits. In my pondering, one thing stuck out to me as I thought about support heroes of the 7.22 meta and the numerous patches before it: the presence of GPM or XP gain talents on popular support heroes. Now, there are always exceptions, but almost every hero that is consistently played in a support role has a significant GPM or XP gain talent in their level 10 or 15 talents. These talents’ value is perceived to be so high that most of them are being taken in 65-80% of the games those heroes play. Shadow Shaman players take XP gain over health 68.1% of the time. Ancient Apparition players take GPM over spell amp 79.1% of the time. Crystal Maiden players take GPM over mana cost and loss reduction 81.5% of the time. I could go on.
This got me asking why this is. Obviously the scaling benefits of these talents feel like they can make a greater impact, as you are always going to perceive your game as having the potential to go longer than it might. Plus, gold and experience are often perceived to outweigh benefits that have tangible stat lines to them like “200 health” and “10% spell amp” because if you want those things, you can just buy it with the gold you’re getting. Now I can probably go into the details of why those might be logical fallacies, and how a significant number of these “economy” talents actually boast worse win rates than their alternatives or even the inverse arguments that these talents are only seeing worse win rates due to a selection bias of people already winning, but I’m not exactly interested in that. What I am interested in, is why these are necessary to exist in the first place, and why players perceive their value to be so high, and that requires backing up a bit.
Power Creep
Before joining the Dotabuff team, I worked as a Game Designer, and one important concept that was taught to me very early on is to be wary of power creep. The slow increase in power of content added to games as their lifespans grow. This is talked about extensively in circles of TCG’s like Magic: The Gathering, but exists blatantly in other games, such as MMO’s that continually release new equipment or poorly balanced strategy games. Power creep is a tough beast to tame, because adding power to a game can be a convenient fix for things that are imbalanced, or for metas that are stale. Adding pieces that are 1% more powerful than what users had before can help keep the game fresh, and renew players' interest in experimenting with strategies, counters, and general theory.
So why bring this up? I think Dota 2 is a perfectly good exemplar of a game that managed its power creep well in terms of its hero design. An inexperienced designer handling the balance of Dota would inevitably patch meta problems by adding more power to the game, then adding more to fix those problems, and eventually balloon the game into a strange, uncomfortable state. Valve and Icefrog however, are some smart cookies and know damn well how to maintain a good budget of comparative strength in hero design. If you look through changelogs, you’ll actually see spells on heroes from 7-8 years ago are mostly similar. In patch 6.72 in 2011, Queen of Pain had her Sonic Wave damage increased to 350/475/600. Today, in 2019, it deals 340/430/520. In patch 6.75, in 2012, Lion’s Finger of Death was buffed to do 500/650/860 damage. Today it deals 600/725/850. I could find many more examples, but the point being, that there haven’t been any significant changes to the Dota balance meta that are worth bringing up… except one. The map and item economy.
As the years have progressed, Valve has slowly, and perhaps inadvertently added more gold and experience to the map, while at the same time, uncapping player’s abilities to harness and use those resources. Does the following screenshot give you any clue as to where I’m going with this?
Item Slots
Item slots were, at one point, a highly valued resource. Not only was there no backpack, but TP scrolls had to be held in the main inventory, and observer and sentry wards stacked separately, taking a slot each. A well-equipped support only really had a single free item slot once their boots and wand accounted for two of their slots. Though this impacted supports almost disproportionately due to the aforementioned wards, this was a major pain point for cores as well. Limiting item slots meant that cores had to make very deliberate decisions about how to upgrade items to maximize those slots. Items with multiple major components like Mjollnir often had to be built outright, or one component like the hyperstone may have to sit in stash while you work on the recipe. This was especially true later in games when a core hero had 3-4 item slots filled already. This resulted often in cores having stagnant power plateaus through the game, where other heroes could catch up in relative power, even if that plateaued hero’s net worth was growing. Gold had to be spent in chunks, and power came in spikes.
In addition to the above slot management, the absolute cap that a core hero could reach was lower. The backpack allows for items like Refresher Orb, Black King Bar, and Necronomicon to be replaced with more stat efficient items once their activations have served their purpose. Valve has also allowed a few key mechanics to escape the bounds of item slots, namely Town Portal Scrolls, Moon Shards, and Aghanim’s Blessings. Boots of Travel’s value used to exist in freeing up an item slot so you could buy an item that gave far more than your power treads or phase boots, and until they decoupled the cooldown from your actual TP scroll, BoT’s saw a massive decline in purchases.
Map Economy
I know I’m hitting a lot of points here, but bear with me, I promise I’m going to make a point of all of this. The map economy has, similarly to the item capacity, inflated. Some of you may remember the monumental 7.00 patch. The patch that saw a massive overhaul to nearly everything, but key among the changes was the addition of new jungle camps on the opposite side of each team’s usual jungle. At first these were balanced with adjusted spawn timings, but as the next few patches rolled in, we saw some of those timing changes reverted, and we eventually settled on the current map layout featuring three camps in a safe “triangle” on either side of the map, and six camps throughout the main jungle.
Patch 7.00 also introduced a new mechanic unheard of before, bounty runes. These also saw multiple iterations in their execution, spawning at different times, offering different mixes of gold and exp at varying spawn frequencies and locations. Regardless of the execution, these served as added gold to the game. They were placed as objectives for players to clash over, but they ultimately added scaling to the game’s economy that wouldn’t have been there before.
In addition to the above two points, a number of other changes affected general farming rates such as the addition of shrines to stay healthy on the map, and a near-global movement speed buff to melee heroes in patch 7.21 which can serve to help heroes move from camp to camp more efficiently for quicker jungle clear times.
Conclusions
All of the above points factor into one main conclusion, that there is not only more gold in Dota, but a greater capacity to utilize it. To confirm this, I sampled average GPM, XPM and Last hit stats from each of the 5 previous Internationals. The results were as follows:
- TI5 - 371 avg GPM, 382 avg XPM, 153 avg LH
- TI6 - 393 avg GPM, 384 avg XPM, 163 avg LH
- TI7 - 383 avg GPM, 432 avg XPM, 182 avg LH
- TI8 - 404 avg GPM, 453 avg XPM, 174 avg LH
- TI9 - 435 avg GPM, 503 avg XPM, 207 avg LH
Keep in mind that the 7.00 patch I cite where a lot of these mechanics were added was shortly following The International 2016, which is where we start to see the averages soar upwards. Additionally, a small amount of inflation in these averages is likely to occur simply due to players getting better and understanding how to rotate through their farming patterns, though I would speculatively say that this doesn’t account for all of it.
The point of all of this is that Valve has slowly unleashed a greater potential for cores to excel in games. I believe a lot of these balance changes existed to help bring games to a close, where late game stalemates in earlier patches would end up dry and uninteresting because teams lacked objectives to fight over, and farming hit its effective cap too early. This has had a knock-on effect though of making supports drop off in effectiveness sooner in games.
Previously, supports had the largest relative impact on a game through the first 10-20 minutes of a game, then gradually, cores surpassed them and had the largest relative impact beyond that. The point where the two crosses over have slowly been pulled earlier and earlier in the game, and I believe this is why we see supports favoring talents that boost their personal economy. Cores are just able to rake in more gold quicker and use greater flexibility in item slots to hold onto more low-cost stat items than they could when they were bound by slot limitations. This makes supports feel out-scaled and less impactful in games, resulting in what I perceive to be a less enjoyable experience. That 600 finger of death damage I talked about earlier, feels far weaker if you’re using it on a player with two bracers in their inventory. Since role queue didn’t exist, I don’t have any way to verify that support players are in lower supply than before, but I know I personally used to find supporting to be a fun and rewarding role, and that luster has been buffered away as time has passed.
I believe Valve is aware of this though. In the last two “letter” patches, we’ve seen neutral creep bounties and exp reduced, and in patch 7.22 we saw death costs reduced, with killstreak bounties increased. There does seem to be an effort, despite it being low-and-slow, to scale back the in-game economy.
Balancing a game as complex as Dota is a feat of epic proportions that takes a number of highly thoughtful individuals to address. Though there tend to be missteps along the way, surfacing the trends that come from these changes can help to inform the future and keep our game the tight, competitive, and rewarding experience it is. I would personally love to see cores have a slower growth rate, but maintain the expanded late-game cap to help close late-game scenarios. I feel supports would have more control over the outcome of a game, and players may find more enjoyment in queueing for those roles, helping to alleviate a lot of the problems we’re seeing with queues. I’m certainly not advocating for a return to a 6 item slot dota, however, I do believe that these changes which appeared to be quality of life-related, ended up having economical ramifications which may need to be compensated for. What are your thoughts?
Really makes you 🤔
I feel support is more enjoyable to be honest. Cheaper obs and sents you can stack, cheaper courier, free upgrade (used to cost you a boots of speed), etc etc. It does create a better QoL than the boots + wand they used to be
However, on the other hand I do agree that there should be better alternatives to GPM XPM talent just like they manage to find better alt to reduce respawn time. What is it, unfortunately, I don’t know yet. They’re not broken, just can be better
I'd go out a limb and say that most of inflation in average GPM and XPM is due to improvements in the economy of supports. Years ago it was pretty normal for a pos 5 to be around level 4 past the 10 minute mark and have only brown boots and a wand to their name 25 minutes into the game. Nowadays either generally means your game is going pretty badly.
Most of the changes in economy seem to me a part of the program to incentivize players in pubs to actually play support roles. Tons of quality of life improvements, lower gap between cores and supports in terms of farm, wider array of support items all contribute to that.
Naturally, cores are also affected by stuff like bounty runes and killstreak gold, but to a much lesser extent, I think.
Weren’t there bounty runes before 7.00? Pretty sure there were
Maybe other thing that differ from the past, now bounty rune affect all team member. This will inflate the overall gpm and improve support's economy
@ETdAWESOME Probably team bounty rune to prevent the 0:00 incident.
I used to play core back in the days before forced to adapt (and eventually love) playing support. It's like parenting. Back in the days, our parent suffer from declining economy but nowadays it was easier (but that doesn't mean supporting/parenting is easy).
#ETdAWESOME there were, but not in the four new spots (and later becoming team rune)
Yeah but the way Canuck says it is confusing, makes it seem as if bounty runes were introduced in 7.00 itself.
Very good article!
This is a great article with a really in depth look at the current game state. But I have to say, there are a few holes that I think should be addressed. As a support player, I think our scaling tools have been increased almost as much as carries. The "team Bountys" effect a support significantly more than it does a carry, we now get gold for stacked camps taken as well as the XP we already got, denies now give money which means when we pull in lane we gain even more side gold while in lane. While this doesn't sound like a lot, before 7.00 we were relying solely on tick gold, kills, and the occasional creep in the jungle.
I completely agree that the game has accelerated in terms of GPM and XPM. I just don't think supports are in this slump. I would argue that even without GPM/XPM talents, talents have created an environment for supports where they can hit significant power spikes more often than just big items (which are inherently hard for supports to acquire). I feel that the early game acceleration of supports in the current meta has led to a game where the 3,4,5 roles can thrive all the way up to 30 min. Rotating around runes and buying cheaper wards to make space and kill enemy cores.
Honestly, I sometimes even wish the game could slow down a bit. I think support was harder when laning was more meticulous and focused on maximum core efficiency. Those are just my thoughts.
Very interested and well-thought article. It's clear that you've developed these thoughts a while and have nice stats to back it up. I don't know what the devs can do to slow cores down and keep supports competitive with them, but now this'll be on my radar in patches to come. I'm interested to see how this question is played out.
@ETdAWESOME Calling me out. For some reason I thought bounties were introduced in 7.00. I guess it was just the change to the team bounties that had me confused. Bounties before that were kinda just a consolation prize to the midlaner that guessed wrong on the action rune though, so they hardly affected anything before that. :D
Really great article, very interesting perspective. As someone who's played very little DOTA in general, and only for the past few years, this was an interesting read for me. I don't know what DOTA was like pre-7.00, and this put it in perspective a little.
@pnapls One of my long-standing ideas for a passion project, would be to use the custom game tools to create a "Dota Time Machine" mod that allowed players to go back and play on any patch in history. Just use the changelog data from the Dota wiki, and go back one patch at a time reverting all changes and try to recapture previous metas. It's probably a ton of work for very little payoff, especially in patches where the map or items changed, but I think preserving that history is important. My girlfriend just got into Dota in the last two years, and every once in a while I think about the fact that she has never once experienced a cliff-jungling prophet, or a bottle-crowing mid magnus, or iron talons and it really strikes me.
Kudos! A great read, with minor mistakes, that doesn't overly extrapolate.
Been working on drafts that implement these factors since before TI9 -- stacking is more important than ever -- very few teams capitalize on this correctly.
Currently, the best course of action is the support staying away and taking the least possible amount of XP from the core, all while, harassing, securing ranged creeps and stacking if possible ; which allows the support to get the instant benefit of the gold from stacks and the XP from the lane once the core rotates out to clear the stacks.
This means the drafts need to focus around heroes who can:
#1 easily clear stacks or help achieve that
#2 easily stack camps without moving too much or stack multiple camps at once
Magnus is a prime example of how the game has changed.
He has been considered a p2-3 since the DotA days, but now he's p4 even p5 for exactly that reason.
Its a terrible idea since he brings literally nothing to the table except empower and needs to stick like glue to a core to provide it.
He's much better off as a real p1 that is self-sufficient ; he clears camps extremely fast, provides that benefit to any other melee core and doesn't want to be the main initiator anyways.
This also means he can solo RP the important targets -- honestly great against those filthy Brood/Lycan/Visage pickers.
KingeY
somebody explain the screenshot to me
Jungling is something many will miss, whenever I think about old dota, that iron talon was the game for both supports and cores. Now, the focus is the laning, so many skills were being adjusted to make laning easier for both supports and cores. Roaming is not as common as before, so you are kind of imprisoned in your own lane until the 5 min mark where you are forced to contest for the rune. I suggest valve bring back jungling and that will bring back more old players. Currently, I feel that options are dwindling, even with those talents, there is really nothing to choose from.
Patch suggestion:
a) backpack renamed utility belt; can only contain tangos, shared tangos, mangoes, clarities, salves, dust, smoke, wards, and TP
b) TP no longer has its own slot
b) TP and BoTs back to sharing cooldown, but CD is 10s lower if it was BoTs that were used (i.e., BoTs 10s CD less than TP).
I also had an idea similar to the guy above except keep the backpack and dedicated scroll slot but include one or two more side slots for utility items like wards, smokes but also blink daggers (basically items that give no stat benefits)
Lol "skills not being changed" meanwhile heroes are literally getting their skills combined and adding a brand new skill into the slot