The specifications are changing much faster than i had thought in this space. I have implemented MVP for my project, and by the time I had completed - the community was discussing Plasma Cash. I feel this is pretty late to ask this question, but still...
What are the real advantages of Plasma Cash? I see here that Karl mentions the main benefits. However i feel the costs seem to be more than just Large token denominations.
Doesn't it cost us the following?
- Micropayments. One time small payments become hard or impossible. Especially where probablistic payments aren't relevant.
- Requires a change exchange. A large entity who has a lot of coins of lot of different denominations. This entity could charge aribtrarily to return change. It feels like a cost to need a third party to make a payment - especially for offline payments.
- Moving a lot of computation & storage to clients. This might sound like a BCH argument, but wouldn't most of the nodes performing the validations be sufficiently large computation devices (laptops, servers)? For low end devices (think India, Southeast Asia, Africa), a few MB of storage is pretty uphill, especially if you are receiving a lot of low value transactions that have changed hands hundreds or thousands of times. Even if the monitoring is delegated to a third party, the exits seem to be costlier than MVP
In contrast, the MVP especially because of the UTXO model didn't seem to have this complication. It had a confirm message requirement which added some complexity, but can be overcome by requiring a maximum block number for inclusion.
I'm still trying to understand what is leading to the movement to Plasma Cash, especially since Joseph Poon 'declared' on a Plasma Implementers call that Plasma cash is now the new Plasma.
Meta questions :
- Is Plasma Cash a true replacement to Plasma?
- What are the usecases that either are more suitable for?
Is there something fundemental i'm missing? I felt it was too late to ask this question on ethresear.ch, apologies! :(