The International playoffs have started and the meta is slowly solidifying. There are definitely some ideas that are slowly being discarded, while other heroes are only gaining in popularity. Today we want to have a look at the outliers — the heroes who are both popular and either very successful or seemingly weak.
Position four Sniper is a poster boy for a bad pub pick that does close to nothing as the game progresses. Turns out pubs were right all along and even in the highest level of play the hero isn’t particularly good. In fact, he is by far the worst popular pick of the tournament.
We genuinely don’t know how the idea even started. Ever since the changes to Aghanim’s Scepter on the hero that reduced the stun duration to almost irrelevant duration, the hero lost most appeal as a support hero. Don’t get us wrong — the Scattershot facet does make up for it damage-wise in the early game, but what about past the laning stage?
The hero offers close to no utility, his damage quickly becomes irrelevant after the 10-minute mark and what you are left with is a hero that can provide some brief vision, apply a non-impactful slow and has a well-telegraphed ~650 damage nuke at best.
The surprising bit isn’t why the hero is losing most of his games, that should be obvious. The surprising part is why professional players with a lot of money on the line decide to go for this option. It is a flex pick, but when you are flexing between a terrible support and a sub-par core, it’s not really that much of a flex.
Now Mirana, on the other hand, is simply insane. Similar to Sniper she is a flex pick, but here the choice is between a very strong core and a decent enough support, provided you have setups for her.
The combined bonuses she gets from Leap, Solar Flare and the level 20 Crit talent make her an incredibly powerful right-clicking machine, while the state of 
Gleipnir make up for her somewhat low farming speed. In midgame the hero essentially gets +160 Attack Damage and +350 Attack Speed, while under the effect of her ultimate.
It is really hard to argue against such values and even though she can’t boast good Attribute gain, she is still an Universal hero, so she also naturally scales well.
Another big outlier with a very respectable 60% win rate. Kunkka lost his waterpark functionality, but he still offers one of the best protections from burst damage as well as one of the best ways to catch people out. Moreover, the hero is extremely flexible in his itemisation, so can adapt to most situations.
We feel like the last part is especially important in the professional scene. There aren’t many mid lane matchups Kunkka loses outright and later on he can go for “silver-bullet” type of items, as his natural Utility and Damage output are high enough on their own.
Dealing with elusive targets? Get 
Orchid Malevolence. A lot of uncontrollable damage on the enemy team? 
Blade Mail is your friend. His utility and damage scaling are also strong enough that you are comfortable going late with the hero.
Dragon Knight being so low in terms of winrate surprises us. We feel like the hero is solid and is more or less a 
Sven replacement in the carry position. He has good laning stage, decent farming potential and scales through having an economic advantage over the enemy.
Perhaps his straightforward approach to fights is to blame. The hero gets in, stuns a target and starts hitting it. This is more or less the extent of his power and he can’t even burst targets down quickly. His tankiness is also a lot less of a problem in a meta where a hero can get 200+ Damage and Attack Speed with a press of a button.
All in all we feel like the hero is not necessarily weak and should be a good pub pick, but being straightforward he doesn’t stand out and doesn’t provide a unique way to play the game — something most core heroes currently do.
Finally, there is Doom who, in our opinion, suffers from an innate that forces players into making bad macro decisions. At the highest level of play the game is all about tempo and having a hero who can’t reliably win his lane, but is dependent on having higher level than the enemy is forcing players into items like 
Hand of Midas, which just don’t fit the overall meta.
When Doom looks good, he looks devastating and unstoppable, but making him look good is hard. It is a bit of a win-more hero, who doesn’t convert a bad situation into a good one, but rather makes a comfortable game even more comfortable.
If the initial conditions of having a good game are not met, the hero is more or less forced into a constant recovery mode and by the time he comes online, it is usually too late. We firmly believe there is a way to make Doom work and it probably involves playing him as a very aggressive and tempo-setting offlaner, rather than a farming one.
This TI isn’t particularly impressive when it comes to hero diversity and overall meta. A lot of heroes don’t work at all and are completely ignored. A lot of heroes are overused because their numbers are a bit too high and a lot of the meta is defined by the best items, rather than overall strategy and unique ideas.
However, there are still several days to go and we hope that the best teams of the International have something up their sleeve to really amp the spectator experience.