[emirozer]

Sorry but this is a bit dumb. If you say YES to this question: "Can you use an always online trusted third party?" Regardless of your other answers, it always says NO as a result.

Can someone give me an example of an always online trusted third party, and also let me know that how can you see into the future so that this third party keeps its guarantees throughout life?

Edit: flagging this post as it seems like a troll attempt to me

[_wmd]

It's little surprise to find someone billing themselves as a 'distributed systems engineer' disliking a tool that helps regular people choose simple designs where appropriate.

From the perspective of a regular user, many "always only trusted third parties" exist - in the western world, for all intents and purposes, they include their utility suppliers, banks, telephone companies, email providers, and chat services.

This site is perfect for someone like my sister who is constantly forwarding blockchain-related junk to me, because her (non technical) job brought her to a few conferences about such things. Yet she doesn't understand the design tradeoffs of such an approach, she only sees the buzz.

[Certhas]

Hey, this is relative to your needs obviously. An always online trusted third party could be the arxiv server for papers. Or a consortiuum of industry players that don't trust each other and set up a foundation that runs a central database. Or a central bank. Really depends on the problem domain.

And yes, I have talked to people who have proposed to use block chain for storing papers. So while it might be obvious to you that blockchain is not a solution for that, it's still a good paper that lays out in detail what a blackchain does and doesn't do.

https://eprint.iacr.org/2017/375.pdf

[dsacco]

>Can someone give me an example of an always online trusted third party, and also let me know that how can you see into the future so that this third party keeps its guarantees throughout life?

This question reduces to, “Can someone give me an example of a third party I’ll trust?”

In general, it is correct to say that if you can trust a third party, you do not need a blockchain. “Always online” can refer to a variey of tiers of certainty. For example, it can mean infrastructure built on AWS or GCP, which is probably safe for the foreseeable future, excepting societal collapse or financial catastrophe. On the other hand, it sounds like you’re interpreting this literally (or at least, far more strongly), such that you require near epistemological certainty that a third party is both trusted and always available (more simply, will never be Byzantine).

The authors are not wrong to discourage use of a blockchain if you can trust a third party, because trusting a third party is simply easier. In the abstract, trusted third parties alleviate the requirement for decentralization and permissionlessness, but different parties have different (and nuanced) risk and trust models. Whether or not you should trust a third party is a function of the value of your data and the perceived resources and incentives of the vendor.

Circling back to your comment - your question is ill-posed, because your implicit requirements for a trust guarantee are likely to be significantly higher than others. For example, I backup my data to Backblaze B2 and Google GCP. Neither of them are “always online” in the literal sense, nor decentralized, so I’m trusting them in particular. Theoretically, backing up my data to a blockchain would be better for thermodynamic trust guarantees based on a distributed, mathematically hard, economically incentivized proof of work. But I don’t need that.

Trust and availability are not binary concepts. Furthermore, you should assess a third party’s trustworthiness and availability based on the value of your own data, not just their capabilities and resources.

[elmar]

>"Can you use an always online trusted third party?"

Real life example at the moment IOTA is working like that, there is one coordinator that works as trusted party (in my honest opinion they will never be able to shut it, but let's wait and see), another example is byteballs another DAG implementation works with 12 witnesses (at the moment controlled by a single physical entity, but in the future can be distributed)