Blockchain seems like the perfect technology to underpin the internal economies of large, complex games.
Some MMORPGs have internal economies of such complexity that real-life economists are studying them, watching secondary and tertiary markets in futures and securities arise in real time as digital real estate, tools, outfits, weapons and toys are sold and bought for in-game currencies that can be exchanged for real money.
In other cases, more simple games see a way to reward players with ‘gold,’ ‘diamonds’ or other in-game currencies that can be exchanged for dollars, Euros and yen if the player desires.
And in the specific case of competitive esports, there’s a particularly close fit: they need to deliver payments to teams that might play together hours a day without ever meeting each other in real life, and who live all over the world. Salaries, prize money and in-game acquisitions need to be sorted out and delivered to the team in an equitable and transparent manner. Blockchain is inherently and uniquely equipped to do this.
At the same time, gamers are already used to the idea of digital tokenized value - in fact their use of this idea predates the Satoshi blockchain.
(If you’re interested in reading more about this, I wrote a big piece on it for Scorum last week.)
All of that sounds like we’ll be moving everything from Super Mario Kart to No Man’s Sky onto an appropriate general-computing blockchain any day now. Why wouldn’t we?
In a word:
Scalability: the elephant in the room
Hassan nails it: scalability is the elephant in the room, that keeps getting slower as it gets bigger. Gaming suffers even more from blockchain’s scalability nightmare than other applications of the technology, because it’s so dependent on speed.
Hassan lives in South Korea, where they have some of the fastest internet connections in the world.
And his set-up’s no joke either: i5, 8GB, graphics card. If using blockchain slows that down, well, Houston, we have a problem. Because one of the biggest things all gamers care about is speed.
Ordinary mainstream gamers will buy a new processor and graphics card; pro gaming rigs go for thousands of dollars, with components built for something crucial: responsiveness.
(This can be yours for just $4,499.99.)
It’s quickness (ping) - not speed (bandwidth)
We’re talking about the difference between absolute speed and quickness, and to make that clear, let me talk - just for a second, I promise - about the different parts of internet connection speed.
When you run a speed test on something like Ookla, you’re going to see three things: upload, download and ping. Upload and download measure how much information you can get down the wire in a second; higher numbers mean a faster connection.
But this measure of absolute speed is less relevant to gamers than ‘ping’ - latency. (OK, they’re not exactly the same thing, but ping is a rule-of-thumb measure of latency.)
Latency measures the time it takes for a single piece of information to travel across the network to another computer and then return to your computer.
Gamers can’t measure time in blocks of one second: that’s long enough to die in-game several times. Elite gamers can make ten moves in a single second.
At every turn, from processor selection to internet speed and acceleration tips, gamers are choosing quickness over speed.
The blockchain a game uses has to be acceptable to gamers who already constitute that game’s fanbase, and their first question will be Hassan’s: will this slow me down?
Whys should a blockchain slow gaming (or anything else) down?
Why scalability is a problem in the first place
Traditional Proof-of-Work blockchains can be slow because they involve mining - carrying out the computer cycles necessary to earn the right to mint a new block.
And the most popular second-generation blockchain consensus model, Proof-of-Stake, has its own drag factors. PoS blockchains require everyone with a stake to sign off on each new block, so more users means more signatories, means more activity, means less speed and - most crucially for gamers - less quickness.
Daniel Larimer explains the issue clearly:
Costs don’t start adding up until you attempt to scale these systems to the size of VISA or Master Card. If you assume everyone is running a dedicated machine processing 1000 transactions per second at 512 kb per transaction, then each machine will require 64 Megabit upload and download internet connection that can easily sustain that data rate, a hard drive growing at over 2 TB per week, and over 256 GB of RAM and a high end CPU. The cost of such a system and Internet connection is easily $1000 per month or more.
At scale you are talking about a total cost of $12 billion dollars per year to maintain decentralization of 1 million validators. This would assume a minimum transaction fee of $3.60 per transaction just to cover network infrastructure.
That’s going to affect gameplay speeds even if the blockchain isn’t being utilized directly for gameplay elements like shots fired or casualties, and only out-of-gameplay elements like in-game NPC-run stores are running directly on the chain.
Can workarounds achieve the speeds we need? (TL;DR: No.)
Solutions for this include the suite of options Ethereum has developed to solve the inherent problems with its consensus algorithm.
To enable scaling and rapid general computing, Ethereum has created both on- and off-chain computing options that facilitate more responsiveness.
On-chain solutions (called Layer 1 solutions) include Sharding, splitting the main Ethereum chain into sections so that the total amount of computation required for each transaction is reduced.
Off-chain solutions (also known as Layer 2 solutions) include State Channels, Plasma and other options to shunt computing operations off the main Ethereum chain.
Finally, Etherum is addressing its underlying problem by introducing the first iteration of its Casper protocol, which uses Proof-of-Stake rather than Proof-of-Work in an effort to accelerate the Ethereum blockchain.
I personally don’t see any of this working. It’s a band-aid solution at best: I don’t think you can turn a go-cart into a racecar one component at a time. In a way, Ethereum suffers from its own success (which I’m the first to salute): because they already have so many active users, they’re trying to rebuild the plane and fly it at the same time.
I see Layer 1 solutions as unable to accommodate very large user bases. In terms of numbers, we don’t need to be looking at Facebook (2.2bn active users) to see a transaction range that will strip sharding of its utility by making the computations required for the sharding process itself outstrip the technical capacity of the underlying chain.
The data requirements of online games are already swamping traditional data centers, and the vast majority of the world’s 2.2 billion gamers - a billion of whom spend money in-game - aren’t using the blockchain (yet).
As the number of users and the complexity of gaming worlds rises, pure data demand is set to rise by 2021 to nearly ten times its 2016 level - a CAGR of 57%.
A system that’s racing to keep up with current demand with workarounds, isn’t going to be in place in three years to handle over three times as much data.
If the billion in-game transactions a month that took place in 2017 were on the blockchain, they would equate to 385 transactions per second - enough to slot comfortably into the Visa network, but 25 times too many for the Ethereum blockchain to cope with. I don’t see how sharding can make a blockchain have 25 times as much capacity.
I’m also not convinced by the potential of second-layer or off-chain solutions. These move computation off the blockchain - but in a fast-moving environment where a purchase or acquisition right now could affect gameplay immediately, and the item in question could be worth a significant amount of real-life money, it seems that this evasion of the ledger is the very last thing we want to do.
Finally, transitioning over to PoS seems unlikely to resolve the critical issue. Both PoS and PoW blockchains scale badly as an unavoidable consequence of their consensus algorithms.
So what should the solution look like?
A blockchain built for speed from the ground up
If significant amounts of gaming switch over to blockchain, they have to solve speed, heaviness - in a word, scalability. Using blockchain technology has to not slow the game down, for everyone or for individual users, or it will be rejected by gamers. And no gaming company will want to alienate its user base.
An ideal blockchain for gaming use would be built from the ground up for speed, scalability and ease of adoption.
That speaks to a need for an underlying consensus algorithm - the heart of a blockchain - that scales seamlessly, eliminating the need for workarounds. The best choice here is Delegated Proof-of-Stake. DPoS blockchains don’t require mining like PoW so they don’t pointlessly eat computing power. And they don’t require an ever-expanding number of nodes to sign off on each block, so more users =/= slower chain.
How DPoS solves scalability and centralization simultaneously
In DPoS blockchains, all users vote for nodes to oversee the minting of new blocks. Thus their voting rights are at one remove - delegated.
Major blockchains that use DPoS consensus include BitShares, Steemit and EOS - as well as its forks, such as genEOS, which I’m proud to say OpenLedger is partnering with. These chains use just 21 nodes to verify blocks, regardless of how many users they have.
Using a smaller number of nodes and delegating consensus power seems like it’s concentrating power in the hands of a few users, making the chain more centralized and more likely to be prey to fraud and manipulation. But a closer look reveals that the opposite is the case.
The current state of decentralization in the EOS blockchain is summed up as follows by Hugo Campanella, from the HKEOS project:
Hugo feels that there’s more work to be done to achieve maximal decentralization on the EOS blockchain, and that as user numbers grow so will decentralization.
This is in contrast to Proof-of-Work blockchains where the economics of mining favor both large capital accumulation and low energy prices, resulting in the concentration of 75% of BitCoin’s mining power in China (and as of 2017, according to Adam Greenfield, 51% of mining power on the BitCoin blockchain in the hands of two companies in China, both owned by the same parent corporation.)
Daniel Larimer, the visionary behind the creation of the DPoS consensus, argues that it provides the greatest possible real-world decentralization:
If you have 1 validator producing 99% of the blocks and 1000 validators producing the other 1% then have you really achieved decentralization? What we really care about is the ratio of validators to blocks produced. Under DPOS we have the ideal ratio, 101 validators each producing about 1% of the blocks.
(In the genEOS chain that’s 21 validators each producing about 5% of the blocks; the principle remains the same.)
Using this system, then, we can build Turing-complete applications on a blockchain that doesn’t slow as it grows, meaning games and gamers can access the inherent advantages of the technology - immutable, distributed auditable ledgers of actions and ownership - without sacrificing the one thing they care about most: speed.
Without this crucial innovation, I don’t see how older blockchains can overcome their scalability problems. But with it, as Moore’s Law drives performance up and cost down on the hardware that performs computations, the scalability lock that keeps blockchain technology from being adopted by games can be picked.
Comments