Because it can suffer the same problem WhatsApp had, because it's centralized. If something on their infrastructure goes down, it might have consequences for all users. This is not the case for distributes solutions like Matrix. (I actually use Signal, though, and not Matrix, because the latter, although it has many advantages, has a relatively bad UX compared to Signal, or at least used to, and it's hard to get contacts to get to use it...)
Centralized management does not mean centralized infrastructure. Any of these systems could distribute the APIs and message delivery infra across multiple independent jurisdictions, regions, and providers, and do clientside load balancing.
The fact that these systems are centralized in management does not mean that they need to have any SPOFs.
Also, the Apple App Store and iOS activation servers are the biggest SPOFs of all. Without them, even with a brand new working device, you can't activate it or install apps to communicate.
> Centralized management does not mean centralized infrastructure. Any of these systems could distribute the APIs and message delivery infra across multiple independent jurisdictions, regions, and providers, and do clientside load balancing.
While true, Signal hasn't done this as far as I know. Until they do, their infrastructure is just as centralized as WhatsApps.
WhatsApp's infrastructure is not centralized; I would bet that this outage was caused by an engineering mistake, not by the failure of some datacenter or cable.
They are owned/managed by the company that popularized "move fast and break things" (though to be clear I do not think that they intentionally take this approach with WA or IG or the FB advertising infrastructure).
> WhatsApp's infrastructure is not centralized; I would bet that this outage was caused by an engineering mistake, not by the failure of some datacenter or cable.
If one engineering mistake can take it down, it's not decentralized. Or an action, not necessarily a mistake. It's at least somewhat centralized. An individual can be asked to take it down. In a correctly decentralized service, that can't happen.
In fact, this did happen last year[1]. Moxie has infamously made the argument that decentralisation reduces reliability because if any one server goes down reliability is affected (ignoring that partial outages are better than total outages).
Interesting. I think it's a semantical battle between reliability (literally "can I rely on the system to be up ?") and resiliency ("if it takes a hit can I keep using it in degraded mode")