gilbert on Warpcast

Stephan pfp

people talk about solana ux but every time i've wanted to swap in phantom i've had transactions fail 5+ times before going through. is this better or worse ux than paying $50 for a swap when the chain is under high load?

10 replies

4 recasts

36 reactions

Chase Sommer pfp

Not trying to disagree to disagree, but I never had a txn fail via Phantom. I have heard of it being frustrating for a bunch of people though.

1 reply

0 recast

2 reactions

Stephan pfp

I’m not a frequent user. Maybe a coincidence with in-app swap in phantom?

1 reply

0 recast

0 reaction

shazow pfp

The transaction pricing/contention is per contract account, so anything that is popular rn should result in failed txns unless the slippage is very generous. But also during peak usage you can always go use an obscure contract and have no problems. That's why peoples' experience varies.

3 replies

0 recast

2 reactions

Apex777 pfp

Wait what. So fees are based on the contract not chain as a whole? Can you ELI5 this in Ethereum speak please 😂 277 $DEGEN

2 replies

0 recast

2 reactions

shazow pfp

Once upon a time, Solana didn't believe in fee markets, for years they kept saying transactions should be ~free (charging a tiny fixed fee) and they just gotta scale harder instead. Turns out ~free transactions means ~free spam/DDoS. After Ethereum did EIP-1559, Solana stopped gaslighting about how dumb fee markets are and introduced a version of EIP-1559 but per account rather than globally for the whole chain. So rather than the whole chain becoming expensive when it's being overloaded, only transactions to the "hot spot" accounts get expensive. On the one hand, it's more granular. On the other hand, it's much less expensive to overload a specific account and put it in a state where you can basically control which transactions get through so you can extract more MEV (especially since a lot of Solana transactions use relayers to get through).

1 reply

0 recast

4 reactions

Apex777 pfp

That's crazy. Did not know that at all. THank you very much. Quick Q, does the contracts for dex's not get hammered too then, or is literally just that token contract? 300 $DEGEN

2 replies

0 recast

2 reactions

shazow pfp

Part of the narrative of this design is that it's up to the contract to shard and load balance accounts so they don't get impacted by load concentration. In the case of a dex, each token pair could be a separate contract. But even thats not enough (top few token pairs get the vast majority of volume etc). If you take it to the logical conclusion, it kinda starts to look like appchain rollups, where each rollup has their own metering.

1 reply

0 recast

2 reactions

agrimony↑🎩 pfp

agrimony↑🎩

yeah i like their local fee market design. would be nice if eth could adopt that tbh. but otherwise isnt solana still using some rng to see which txs land? or is that design now removed

1 reply

0 recast

0 reaction

shazow pfp

I support multi-dimensional fee markets, and even scaling costs per-opcode, but I think local fee markets are worse for decentralization. The entire premise of economic security is to offer uniform properties to the entire surface of the chain. If it costs a tiny fraction of the cost to attack a specific contract, then why even be a member of the most secure chain rather than your own chain? May as well use Cosmos.

2 replies

0 recast

1 reaction

agrimony↑🎩 pfp

agrimony↑🎩

interesting thought. hadnt considered the cost of dosing the contract. how does solana get away with it, or they simply just dont care about decentralization?

1 reply

0 recast

0 reaction

shazow pfp

"why are all my transactions failing?" ($TRUMP buyer) "i don't know what you're talking about, *my* transactions are perfect, clearly you're doing something wrong" ($OBSCURECOIN buyer) is how solana gets away with it :)

2 replies

0 recast

2 reactions

gilbert pfp

The fun part is all accounts are required to be explicit in any Solana tx. Probably makes the life of an MEV author pretty easy, maybe even adding new possibilities

0 reply

0 recast

3 reactions