DEXes do not have to be expensive at all. This is just a side effect of most people building the wrong stuff on the wrong "blockchain".
It seem quite obvious to me that a DEX should be somewhere where there are cheap and fast transactions.
> It seem quite obvious to me that a DEX should be somewhere where there are cheap and fast transactions.
Interesting, I would have thought that Centralized exchanges would be cheaper and faster.
Cheaper because there are no block chain transaction fees. So assuming a DEX and CEX can charge the same fee structure, then CEX's should win as there is no exchange fee added to each transaction.
On chains like ETH that matters alot if you trade alot.
And same for speed. On a CEX a trade is a database update, on a DEX its a blockchain update. Hard to see how the later can be as fast as the former.
Why should a centralized exchange be cheaper? There is always a middlemen who removes value from the system.
>...then CEX's should win as there is no exchange fee added to each transaction.
The other way around, the DEX should win as there is no fee at all.
If you could make a decentral Ebay there would be no Ebay to take a part of your profit.
The problem with transaction fees in many blochchains is that they actually replaced the middlemen with a decentralized middlemen(s) (the miners/stakes/node operators etc.) Now you have a system operated by people who want high fees and user who want low fees.
They operators cant really collude because they compete but they know low fees would hurt them all so they basically form an oligopoly without even talking with each other.
Ideal the system should be operated by the user so they would all want fees to be low.
I dont know if any such system exists but I know the XRPL solved the problem by burning the fees so node operator have no incentive for high fees. This results in a typical Tx cost of less than $0.0001 USD. Any offer on the DEX is just a special transaction and does not cost anything extra.
The fee is only a negative incentive so people dont spam transaction.
>And same for speed. On a CEX a trade is a database update, on a DEX its a blockchain update. Hard to see how the later can be as fast as the former.
This is true a CEX will definitely be faster with current tech and I dont know if that will ever change but for manual trading this is not really a problem. XRPL DEX executes orders about every 3-4 seconds and the consensus mechanism does not allow front-running.
> Cheaper because there are no block chain transaction fees. So assuming a DEX and CEX can charge the same fee structure, then CEX's should win as there is no exchange fee added to each transaction.
Sorry I meant block chain transaction fee, not exchange fee, I mistyped the second instance.
> The problem with transaction fees in many blochchains is that they actually replaced the middlemen with a decentralized middlemen(s) (the miners/stakes/node operators etc.) Now you have a system operated by people who want high fees and user who want low fees. They operators cant really collude because they compete but they know low fees would hurt them all so they basically form an oligopoly without even talking with each other.
Trustless distributed consensus is expensive - running chain validation isn't free, and there's a tradeoff where the cheaper you make it the easier it becomes to attack. That's why this stuff ends up costing more than a traditional middleman who can use a cheaper datastore and consenus mechanism.
Consensus is actually very very cheap if its not done with PoW or similar consensus "tech".
Distribution isn't cheap because obviously 2 identical servers cost twice as much as one. But this comparison isn't actually applicable to real world use. A centralized system that wants to have high uptime and reliability does need multiple server too. Most centralized system do not reach the kind of reliability "blockchains" can have because from a cost-benefit point of view it just does not make sense. 99.999% uptime or about 5.5 minutes downtime per year is very expensive and very few public system come anywhere close to this.
>...there's a tradeoff where the cheaper you make it the easier it becomes to attack
That is actually a pseudo-math-security coming form the early days of Bitcoin.
Yes, if you double the number of nodes the system is arguably more secure (if the nodes are actually operated by different entities) and it costs twice as much to run BUT there is no point to endlessly increase the number of nodes. Less nodes but strategically placed around the world operated by different entities is way way more efficient and cost way less than the blind "more is better" approach.
Some quick back-of-the-envelope calculation:
1 node hardware + energy and maintenance cost per year $100k USD.
50 such severs distributed all over the world.
This system would therefore cost about 5 million per year.
Sounds like a lot but assume there are 1 million transactions per day that would mean 1 Tx cost only $0.0137.
Obviously more Tx/d would reduce the price per Tx because operation cost cost does not grow at the same rate.
Give $0.0137 USD to the operators for each Tx sound like a good idea but most system also give the operators the power to change that price, which is comically stupid.
The reason Tx fees on some blockchains are order of magnitudes higher, is because of bad design, completely pointless levels of redundancy, and most importantly because the operators do make huge profits.
Its only logical that they would double the fee if that reduced the number of Tx by less than 50%. They dont care if you cant use the system anymore because the fees are too high for you they only care about the sum of all fees.
Imagine HN but every comment would cost money and the operator would want to make the maximum profit possible.
You could plot the average number of comment at a given cost and find the price that leads to the maximum profit.
One thing is for sure, comments would not be cheap.
Have a look at https://news.ycombinator.com/item?id=33536736
There are ofc other solutions where L1 is cheap, fast and secure, Ethereum is just really not the right tech for a DEX.
