This is disappointing. What's been read can never be un-read; to say otherwise is deceptive.
This is disappointing. What's been read can never be un-read; to say otherwise is deceptive.
In this case, yes, it's impossible to guarantee that some malicious peer doesn't ignore my "plea to delete". But combined with the fact that my data will only be replicated to/by peers I already have a trust relationship with (as opposed to e.g. on a blockchain) it provides another layer of protection that a system without deletion simply doesn't have. Not perfect, but not useless either.
The project's goals are hard and noble. It would be better to under-promise and over-deliver than to make everyone question their claims. Maybe I'm just a grumpy old man at this point, but there are already too many caveat emptors in computing. They could have said "better" erasure of data.
You're right that it's nigh-meaningless for a public cluster.
Attackers are a whole other matter, and their existence doesn't make the feature pointless, for the above reasons.
This is a good point, and "GDPR compliant erasure of data" would be a great way to explain it. As a user I can guess what that means, and as an engineer it doesn't sound like magic.
I appreciate people trying to do something hard and noble.
Maybe "total erasure of data" is a too strong promise, but the fact that you can not force nodes that you don't control to unsee things is common knowledge, so in my opinion this does not need a qualifier.
Willow's claim has to do with erasure of the _networked_ data. It doesn't claim that copies people make are destroyed. Almost everyone understands and expects that if you can view data, you usually can somehow make some kind of copy of it. The question usually comes down to: how good of a copy?
Perhaps the best way to prevent perfect copying of data is to prevent someone from viewing it on a device they control.
This is true for the target audience of the article, but certainly not for people in general. It might be true for people in their 20s, but I strongly doubt it's true for any other age range.