What differences did each one propose in comparison to the previous network? What was the reason for each testnet to be introduced? When was each test network introduced? The question includes from Testnet 1 to 4 (or even 5) and signet(s).
There are two similar questions 36252 and 9975, but they are old and outdated. I am looking for a detailed historical description and motivation about each test network introduction.
The two linked topics covered Testnet1 and Testnet2. From what I gather: Testnet1 (launched 2010) essentially became the first altcoin when people began trading it. It was abandoned by resetting the ledger to create Testnet2 in February 2011. Testnet2 also started to get traded for money and its difficulty got too high to be useful for testing. In February 2012, Testnet2 was hardforked to introduce the 20-minute exception, which requires that a block with a timestamp at least 20 minutes greater than the preceding block must be mined at difficulty 1. Both Testnet1 and Testnet2 preceded Namecoin which was launched in April 2011.
Testnet3 was started in April 2012. It also incorporated the 20-minute exception just introduced two months prior to Testnet2. This was supposed to make the testnet sufficiently "broken" to stave off monetization and allow any users to mine blocks if the difficulty got run up. Otherwise blocks were required to adhere to the difficulty as calculated per the usual rules. Unfortunately, when this exception is used on the last block of a difficulty period, it causes the difficulty reset to base its calculation on the minimum difficulty which almost completely resets the difficulty. This led to “block storms”: after a reset to almost minimal difficulty, blocks would be mined extremely rapidly until difficulty reset several times to a higher level. Later, attackers exploited the block storm vulnerability to keep the difficulty programmatically low for an extended period throughout which a block storm persisted. The block storms caused Testnet3’s blockchain to proceed much more quickly than intended and exhausted the block subsidy before its time: today, Testnet3 has just under 5 million blocks which should usually take about 95 years instead of 15 years, and after its 23rd halving, the subsidy is 596 sats. Additionally, market places started again to monetize Testnet3 coins, which on one hand made it more straightforward to get some, but on the other hand disincentivized returning them to faucets and faucets from handing them out.
Testnet4 was launched in May 2024. It kept the 20-minute exception, but based the difficulty calculation on the previous period’s first block instead of the last block. Because the first block is exempt from the 20-minute exception, this prevents block storm attacks. However, some users started systematically post-dating their blocks timestamps 20 minutes past the prior block to quickly mine the new subsidy. The popularity of using the 20-minute exception reached an extent that causes a large count of competing chaintips at most heights and frequent chain reorganizations. Some of the same parties monetizing Testnet3 immediately started trying to monopolize and monetize Testnet4 coins.
Screenshot via https://x.com/0xB10C/status/1876642278439346339
Thanks for corrections by Pol Espinasa (block storm vulnerability) and Antoine Poinsot (introduction of 20-minute exception in Testnet2).