Doesn't matter. The moment the canary died, they became a compromised agent whose further statements are to be interpreted as being manipulated by secret court order.
If your canary dies, you can buy a new one, but you still have deadly gas in your mine. You can't trust the words or actions of someone who just declared they have been compromised.
>Doesn't matter. The moment the canary died, they became a compromised agent whose further statements are to be interpreted as being manipulated by secret court order.
I don't think you can draw that conclusion. As I understand it, the NSL can only compell them to shut up about it, but it cannot compell them to dish out arbitrary lies and false justifications for taking the canary down.
We can and should draw that conclusion. Anything less compromises the canary system. I'm not going to bet my security on someone who frivolously shouts fire in a crowded theater, even if they later say they were only kidding.
You can of course do whatever you want as a consequence of what they are saying or doing.
What I'm saying is that it doesn't logically follow from the legal situation that their "further statements are to be interpreted as being manipulated by secret court order".
The FBI cannot arbitrarily manipulate SpiderOak's further statements.
They never declared that if the warrant canary dies they can no longer be trusted in anything they say or do ever again.
We know exactly what they can and cannot legally say or do once an NSL has been served. Therefore we also know that they cannot be legally compelled to make false statements.
You're right, it's more likely they are putting out the statement because they don't want their business to be ruined by having admitted to receiving an NSL. There are many ways someone can be coerced, and I don't need to guess the one in play here to discount their further excuses out of hand. At the point at which the canary is removed, their untrustworthiness is tautological; there is no combination of truth values which result in them being considered trustworthy.
That is the feature of canaries that makes them respected, and the need for such features is why secret courts are so corrosive to our society.
While I agree the canary should be considered dead on principle, I don't find this particular argument very convincing.
I think the main point here needs to be that the entire notion of a canary only works if we can count on them being killed only for the purpose they were created. If it becomes acceptable to kill canaries because the signers are tired of signing them then we have a bit of a problem.
Now, for this specific case...
It sounds like you are arguing that a NSL has compelled them to make false statements. This seems fairly unlikely given what we know about the current legal situation. The idea that they may be lying for the sake of their business is more convincing.
However, if they are lying for whatever reason, why not just continue to sign the canary? The only plausible reason to do this is some sort of malicious compliance with a sloppily worded order compelling them to lie. It would make no sense for them to decide to lie for the sake of the business and then do all of this instead of just signing the canary.
If they have received a NSL this basically leaves two sequences of events, both of which contain some rather unlikely events. If I had to bet, I'd probably bet against them having received a NSL.
However, the canary should still be considered dead. They literally killed the canary. Their reasoning provided for it is really quite bad. Basically they ask users to trust their unsigned website because users already trust their (closed source) code. So they have either received a much more powerful NSL than thought legally possible, or are doing the worlds worst job of lying about not receiving a NSL, or they have not received a NSL and view a regular page on their website as a suitable alternative to their previous solution which involved three people in three different countries signing the canaries with keys stored on air-gapped computers. If you are counting on them for security, none of options are good.
I think there's an argument to be made that the canary continuing to live is not sufficient to establish trust, for the reason you outlined. However, given that it is in fact dead, this is sufficient to revoke trust-- as you also pointed out, none of the scenarios are good for them.
As for why they would kill the canary and then backtrack, I think a plausible story could be that the canary was killed by an engineer, and then the backtracking happened by management because they don't want to own the consequences. This could explain why there was a delay between removing the canary and putting out a cover story.
Collectively as an organization, they're either acting in bad faith for whatever reason, or else they're incompetent.
I mean, a canary is only as good as your trust in the people putting it out.
As for the engineer/management idea,their statement revoking it was signed the same as their canary. Aside from the exceptional circumstances I described earlier, if you trusted their canary then the statement should be trusted too.
Like I said, I think we're most likely looking at incompetence.
If, for the sake of argument, a hostile (to the security of SpiderOak users) government actor was able to arbitrarily manipulate SpiderOak's further statements for purposes of furthering said government agency's goals, what might they have it say and how would those government-sourced statements differ from what we're seeing?
It doesn't have to be direct manipulation, and it doesn't have to further a government agency's goals. It could be a voluntary self-interested act by SpiderOak, and they should still have trust removed.
Oh, I agree that they cannot now be trusted. I was just trying to indirectly make the point that from our POV SpiderOak acknowledging they made a mistake, or them deliberately covering for a NSL look identical.
> The FBI cannot arbitrarily manipulate SpiderOak's further statements.
Maybe not directly, but they could be coerced into it. A sort of "it's in your best interest to redact XYZ.. or else" unwritten statement would do the trick for most folks.
you should never base your decision on something published on the web site. you should rather base your decision on review of open source code, and on-device key management.
Since none of this has been tested in court, the subsequent statements could be insurance on their part to avoid a bad outcome. Conceivably a prosecutor could argue that by removing the canary, they've published one bit of information in violation of the gag order. By publishing a bunch of excuses for their action, they give a jury reason to disagree.
That's a possibility, but it seems rather contrived. Maybe the FBI threatened to challange their canary in the courts unless they came up with some sort of credible excuse for killing it.
If your canary dies, you can buy a new one, but you still have deadly gas in your mine. You can't trust the words or actions of someone who just declared they have been compromised.