The article includes problems regarding the testnet which I didn't know about:
The fact that testnet blocks are mined by volunteers means there are two big trade-offs when it comes to reliability. One is variancy in the block intervals (this may change from seconds to hours or even days); second are the reorgs (a few thousand-block reorgs have happened in its history). The block interval variance mostly represents changes in the difficulty adjustment or miners plugging in and out of the network, changing its hash rate.
I then looked up the hash rate of mainnet and testnet in mempool.space:
mainnet hash rate
testnet hash rate
As you can see, there is indeed a vast difference in consistency and hash rate between mainnet and testnet with currently around 250 EH/s for mainnet and around 1PH/s for testnet.
I asked myself now if it could have a positive impact if more people just let their full node mine on testnet? Could be CPU limited. It wasn't that easy to look up CPU SHA-256 benchmarks since ASICs entered the bitcoin mining scene as early as 2012 according to an article on nicehash.com [1]. So I don't know if this is very far fetched to make a dent into 1 PH/s with (limited) CPUs.
I am at least interested in this so I can learn more about mining and
getblocktemplate and stuff. So maybe I'll just do it regardless what consensus on "let's tell people they should mine on testnet" is.The size of the testnet blockchain at least does not seem to be a problem. First research has revealed the size to be significantly less than on mainnet:
As of January 2018, the size of the data on disk was 14 GB containing data for about 6 years worth of testnet activity.