Fees have risen to $0.075 for a simple transaction that takes less than two minutes, according to EthGasStation, and as high as $0.58.
Smart contract based transactions, like opening or closing a position on MakerDAO’s Dai, can cost more.
In addition token transfers are more expensive than simple transfers, with Tether apparently taking up half of the available gas.
In addition Binance moved to accept only eth based USDT. So if one wants to arbitrage say from Huobi to Binance, then they’ll have to go through eth.
Meaning half of ethereum’s decentralized and global blockchain now serves solely cash like digital dollar fiat, giving miners about 400 eth in fees with 800 eth paid in total by all network users, or circa 7% of the entire miners’ reward at around ◊13,000 a day.
The somewhat decent rate of daily fees at about $150,000 may be one reason why some miners would rather increase them further than lower them.
Ethermine in particular seems to charge the highest rate from known miners. They are voting down the gas limit, together with Nanopool, so preventing the 50% of miners’ vote to increase it.
Ethermine appears to be a big supporter of ProgPoW, a controversial algorithmic change of ethereum’s network, presumably on the basis it would kick asics miners out and they would earn more.
Meaning it may be they’re voting down the gas limit purely in the hope of getting more money, as tech wise orphan (uncle) rates are considerably down:
We can see a gradual decrease over two years and now a somewhat stable, but slightly falling, level in eth uncles.
Meaning technically ethereum can handle about 4x current capacity due to numerous improvements to the main node client, Geth.
Etherminer however might be running Parity, with it unclear just how much they have optimized their software.
On a surface view it looks like miners’ hash and uncles correspond, but Spark Pool seems to be a bit more efficient, while Nanopool seems to be struggling in this regard.
Nanopool has less hash/blocks than F2Pool and yet they seem to get more uncles for some reason, with that explaining their vote down.
Why Ethermine is voting it down is unclear, but a number of further improvements are expected with Istanbul sometime perhaps in November-January.
Then there are the eth 1x changes, some of which apply retroactively, with it unclear at this stage when they might go through, but once they do a number of concerns from storage to node synching should be addressed.
This limit, however, is what can be called a soft limit. Unlike bitcoin’s 1MB which requires the entire network to upgrade, in eth miners can just increase it if 51% of them agree to do so.
Due to pool centralization, 51% usually means 3 pool operators or less, but most of them are actual pools whereby ordinary individuals, farms, and all else in between, join together under one hash.
So miners could just move to a big block pool if they wanted, with the limit here increasable in a gradual manner say to 12 million gas.
At the current 8 million it happens to correspond to exactly 1MB of data every 10 minutes. As bitcoin has long increased theirs to 2MB, arguably eth could easily go to 16 million gas, but whether miners will do so remains to be seen.