Okay, I found the main article I was thinking of when I left my comments above. This is an article I've referred to in previous answers where cryptoeconomics was of interest. (For example, this one.)
There is an article from earlier in 2018 written by ConsenSys that looks at the comparison between Blockchain and DLT in detail:
Blockchain vs. Distributed Ledger Technologies (External link.)
There are several key points of comparison in the article, including the difference in the use of cryptoeconomic incentive layers, which I mentioned.
The TL;DR is as follows:
- "Blockchain" is the label given to those technologies that are operating in an environment where there is little or no trust between transacting parties. In such cases cryptoeconmics comes into play to incentivise participants to remain honest.
- "DLT" is the label given to those technologies operating where there is a level of trust between participants, such as private or consortium networks. In those cases there's no need (or at least less need) to incentivise transacting parties to be honest, and therefore no requirement to incorporate the game theoretical considerations of blockchains. (i.e. Mechanism design is less important.)
From the article:
"A core tenet that distinguishes blockchain ecosystems from
distributed ledger-designed database systems is the capability to use
mechanism design as an economic incentive layer that ensures proper
allocation of trust and cooperation to make a system behave in a way
that is conducive towards instantiating decentralized consensus among
users as well as security."
So, for a blockchain platform such as Ethereum:
"An example of cryptoeconomic incentive layers can also be seen in
Ethereum’s transition to a proof of stake consensus mechanism via
implementations of Casper."
But for a DLT platform, such as R3 or Hyperledger:
"An important note to recognize is that R3 Corda and Hyperledger
Fabric do not have these cryptoeconomic incentive layers instantiated
within their software architectures. Due to the fact that the software
architectures are foundationally designed based on distributed
database focused paradigms, they were not originally designed for the
incorporation of native cryptocurrency layers within the overall
framework."
As Rob mentioned in his answer, this is a complex topic, and my personal interest in cryptoeconomics perhaps skews my opinion of what the most important differences are.
And, finally, while ConsenSys are a well-respected leader in the space, they don't get to define what terms like "blockchain" and "DLT" actually mean. It's likely the World Bank have their own ideas...