4 ms·
Total erasure of data. This is disappointing. What's been read can never be un-read; to say otherwise is deceptive.
by ynniv 3y ago
Total erasure of data.
This is disappointing. What's been read can never be un-read; to say otherwise is deceptive.
- black_puppydog 3y agoIn the same way, once an attacker can exploit some weakness in a system, it's game over. Yet defense in depth is a thing and makes it much less likely that bad things happen. 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.
- ynniv 3y agoYes, but that's not what "Total erasure of data" means. 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.
- wharvle 3y agoIt's a handy feature to have in the protocol if you're operating detached networks using it, and can control all the clients. If you're using it as internal infrastructure. Which, personally, is the only way I've ever been interested in using these sorts of things. You're right that it's nigh-meaningless for a public cluster.
- ynniv 3y agoOk, but where exactly is that when the FBI is pulling Bitcoin private keys out of files accidentally synchronized to iCloud Files? These are hard problems.
- wharvle 3y agoI don't understand your concern. Guaranteed (if you control the clients, for ordinary values of "guaranteed", not, like, mathematically-rigorous ones) deletion is a handy feature if you need to be able to comply with regulations, or just want to be sure you're not wasting disk space on stuff you intended to delete, without having to do extra work. Attackers are a whole other matter, and their existence doesn't make the feature pointless, for the above reasons.
- ynniv 3y agodeletion is a handy feature if you need to be able to comply with regulations 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.
- detourdog 3y agoWhat is the point of noting trying to accomplish hard and noble things. I'm sure there are plenty of people willing to take shortcuts. I appreciate people trying to do something hard and noble.
- ynniv 3y agoMe too! But let's be honest about how things work.
- detourdog 3y agoI thought I was. I don't think it has much to do with the way the world works. I think it might have more to do with how one works the world. There are plenty of people that don't want to try the impossible but the impossible should be explored.
- rklaehn 3y agoPrefix pruning is a very different approach than tombstones. An update will actually remove data, and not just mark the data as removed. 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.
- xpe 3y agoI can appreciate the spirit of the comment. I'm more pedantic than is typical -- perhaps even more than is healthy! Even so, I don't think the claim is deceptive. 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.
- aidenn0 3y ago> Almost everyone understands and expects that if you can view data, you usually can somehow make some kind of copy of it. 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.
- xpe 3y agoI take your point. How well this is understood is an empirical question. I retract my claim that it holds for "almost everyone" from the broader population.
- orthecreedence 3y agoI don't really get what your nitpick is here. There is no conceivable way in the universe to unshare information that has been shared. The idea here is that you can stop further sharing of that information. I think that's a fairly reasonable and obvious interpretation.