3 ms·
I'm glad we switched away from Ethereum in favor of Stellar; it will take most of this year and probably next for the dust to settle on this. Until then, Ether
by jillesvangurp 8y ago
I'm glad we switched away from Ethereum in favor of Stellar; it will take most of this year and probably next for the dust to settle on this.
Until then, Ethereum continues to be unusable for serious dapps (which are rare to begin with) that have expectations that exceed having reasonable transaction throughput (<4/s, globally, for everyone) and latencies (minutes to hours).
Until then, this is a toy platform for wannabe ICOs, misc gambling applications, and not much else that I would qualify as substantial in transaction volume or user base. I
f your use case is "nobody does anything that requires a transaction to happen more often than maybe once a day and you can justify the high tranaction cost", ethereum is for you. Otherwise if say, you expect to have users that do stuff in your platform, use something else that works right now instead of waiting for the next two years for Vitalik et. al to actually deliver on their promises with something that is usable, scalable, stable, secure, and available.
We switched to Stellar, which also implements proof of stake and provides a reasonable compromise between security and availability. It has its own challenges but transactions clear in <4s and there are dozens to hundreds of them per minute globablly. It also skips the convoluted smart contracts (which IMHO are severely expensive, hard to fix, exploitable bugs waiting to happen) in favor of simple but adequate financial constructs builtin to stellar for custom assets, transactions with multiple signees, transactions with limited time validity and many other features that you can combine to emulate many of the more things people attempt to do on top of solidity. E.g. you can do escrow transactions, payment channels, and many other things. It also scales down reasonably well for micro transactions with values measured in a percentage of a cent (though payment channels are probably the way to go for that).
- wslh 8y agoThe Stellar protocol has not been yet peer reviewed. Not saying that is wrong but I would say that when you start relaxing consensus assumptions you can come up with even simpler protocols. For example, using trusted third parties to timestamp your transactions. In this way you solve the ordering issue in consensus (assuming no timestamp can give the same time). For example, imagine if Google, Amazon, Microsoft, and companies from other countries provides this (Microsoft has a timestamp server for creating code certificates).
- akerro 8y ago>For example, imagine if Google, Amazon, Microsoft, and companies from other countries provides this (Microsoft has a timestamp server for creating code certificates). Sooo... paypal on blockchain?
- justtopost 8y agoPlease no. Keep paypal on ebay, and away from the rest of us.
- wslh 8y agoNo, saying that you can rely on third parties just for the timestamp but not for the whole blockchain
- jillesvangurp 8y agoI understand; this has also been a concern for us. However, there's actually very little out there that is actually mature and usable. Bitcoin is not usable for custom applications period. It's nice if you want to trade bitcoin; it doesn't do anything else. Ethereum has enough of a user base that you could argue it is probably OK in terms of security; however as argued it's not really usable in its current form. Everything else is a combination of too young, technically challenged, too small in user base, etc. Bearing this in mind, we picked Stellar based on the fact that it has actually been around for a while and launched late 2017, is used by several more serious players in the ecosystem (e.g. IBM seems to like it) and a few fin tech companies are building stuff with it. It has a few tokens that seem less about speculative trading and more about solving a real problem (e.g. several stable coins that tend to be relatively stable in value, by design). If you look at e.g. EOS, the picture is a bit different. It seems to be favored by asian gambling tokens which have pathetic numbers of transactions per day. Also, they launched after we decided to move to Stellar. Tron seems to have similar issues. Then there are a lot of smaller chains that are currently not launched yet e.g. IOTA, hashgraph, that you might see as potentially of interest in the future. After that comes a long tail of solutions that nobody can vouch for right now. So, we're taking some risks here with Stellar but it seems to be relatively OK for us so far and I haven't really seen anything more suitable for our needs so far. In terms of transactions per second there's way more happening on Stellar than Ethereum is even capable of handling currently. Also, Stellar seem to have distanced themselves a bit from the whole ICO business, which I count as a good thing.
- seibelj 8y agoStellar, like Ripple, is centralized. They run an append-only database that only they can approve transactions on. Ethereum is attempting to be far more decentralized, which is why it’s so difficult. Stellar may solve problems but the types of problems it solves are nothing like what Ethereum is doing. And smart contracts are indeed important for many applications, despite their flaws.
- jillesvangurp 8y agoNot as black and white as that. There are multiple companies running validators and you can choose to run your own. The transactions go to the same shared blockchain. Most of the significant tokens on Stellar have their own preferred validators for transactions. Each validator can configure consensus for their ecosystem as well. IMHO this is a reasonable compromise and not that different from Ethereum proof of stake in combination with a high stakes. Ethereum is a nice science project so far that may one day result in something usable and when that happens, I will look at it again. However, that is not right now and we are trying to run an actual business right now. Most of the smart contracts I've seen are convoluted and tend to solve things that Stellar can do out of the box with zero programming. The smart thing seems to be to reuse smart contracts from Github and avoid writing your own and thus avoid taking the risk of introducing exploitable bugs right in the heart of your financial operation. Not to say that they are useless, but I have no need for them with Stellar. That saved us a lot of time so far.
- snissn 8y agoAs a smart contract developer I'm curious about > It also skips the convoluted smart contracts (which IMHO are severely expensive, hard to fix, exploitable bugs waiting to happen) in favor of simple but adequate financial constructs builtin to stellar for custom assets. To me that feels very very limiting and my optimism about Ethereum is the generality of the smart contracts and ability to programmatically enhance Ethereum.