/biz/ - Business & Finance

>>53531581
YIKES!

The fact that it would be engineering intensive and difficult just to integrate a non-proprietary VM is so fucking bearish given they posture as some sort of world omni-computer. ICP is a relatively complicated system takes a monolithic architectural approach. Generally speaking that is not a successful approach for technology.

1. Marketing speak, ICP only does ~450 TPS per subnet, who knows how much per canister they don't advertise that. Finality is 2-3s which is definitely not "web speed". Frankly embarrassing given the hardware specs of ICP nodes.
2. That falls apart if you exceed the capacity of a single subnet, then you're bottlenecking with cross subnet communication, more marketing BS.
3. Except you're limited by the host chain's performance, which for BTC and ETH is dogshit. You aren't actually bringing assets into the ICP ecosystem with these "integrations".
4. No users having expenses is what is going to bite you in the ass on adoption. Making dapp deployers foot the cost of usage instead of a flat hosting fee like web 2.0 is not an advantage.
5. True if everything else is build on ICP, but it's not and won't be. Also you're going to run into latency issues cross canister/subnet that will make for dogshit UX. You can't just layer a bunch of computationally intensive consensus/cryptography processes on top of normal database functions and get equivalent performance, there's no free lunch.
6. Meaningless marketing drivel.
7. Absolute kek. The hardware requirements for a node are astronomical, meaning only server farms are going to be running the ICP which are definitely not decentralized (no geographical distance isnt the only thing that matters.) The real kicker though is that ICP is explicitly a permissioned consensus network with KYC nodes which is straight up not decentralized in any crypto usage of the term. It's the only reason they get some semblance of performance (which as stated earlier isn't that impressive).

>>53488880

>>50227634