Imo, their algorithms can be tested in developer environments (testing/development servers). And even if they absolutely HAD to do it on the live servers, then they could have easily removed their diggs after testing.
Their's just making an excuse for a mistake. Their way of handling this is understandable, but probably not the best way to handle that situation.
I wanted to write exactly the same comment. I don't buy their explanations. Especially when they write: "we continuously run tests on the site to expose vulnerabilities in our own security." Does that mean they run randomised testing? Basically something that they have no idea what will it do in the end? I don't believe they'd be ok with trying out something that can potentially uncover a data corruption issue in their live system.
More likely, they paid someone to do blackbox security testing on the live site. In that case, they would be pretty lucky if all someone could find, was a way to digg stories multiple times.
Very true. If it was really testing, why would they stop exactly and immediately when informed by the blogger that their behavior was 'caught'? They would've continued the so called experiment if it was really one.
I think they did it to inflate the number of diggs on the front page stories so that it looked like more people were frequenting the site. That's the reason that there's no apparent bias in the selection of media outlets whose articles the accounts digged up.
>If it was really testing, why would they stop exactly and immediately when informed by the blogger
That's absolutely not evidence that it is not testing.
It's a good point, but it is in no way inconsistent with their explanation. It makes perfect sense because it was (theoretically) supposed to be a small test that nobody noticed. Once somebody noticed it, of course they're going to stop it.
It's consistent with other theories being floated around, but it's important to note that it in no way contradicts their statement; exactly the opposite in fact.
If your theory is correct, why not just do it more directly? Why not do it in a less obvious way? Why upvote random spam sites as well (some of which were created at the same time as the accounts in question...almost as if someone were testing submitting spam on Digg...)
I don't doubt it's possible, but it seems to me people are letting their bias against Digg get the better of their reasoning ability.
1) Let's say their excuse is bogus. So what where they doing? Gaming the system like ltgenpanda claims?
There's a contradiction there though. On the one hand you're saying that you don't buy it's testing because they're too smart and too capable to test on production...
...and on the other hand you believe they have to resort to this roundabout way of gaming their own system because they are not capable and smart enough to just do it a more direct way.
2) Your assumption (that they could do it in a test environment) is not necessarily true. If their explanation is to be believed, the testing is not the same as say normal software testing. The intent is to be a realistic security test...more like a "pen test"...which are quite often done on production machines.
So far I haven't seen anything that contradicts their explanation, and indeed it fits the facts as we know them pretty well, while the "game theory" doesn't.
Comments
Imo, their algorithms can be tested in developer environments (testing/development servers). And even if they absolutely HAD to do it on the live servers, then they could have easily removed their diggs after testing.
Their's just making an excuse for a mistake. Their way of handling this is understandable, but probably not the best way to handle that situation.
I wanted to write exactly the same comment. I don't buy their explanations. Especially when they write: "we continuously run tests on the site to expose vulnerabilities in our own security." Does that mean they run randomised testing? Basically something that they have no idea what will it do in the end? I don't believe they'd be ok with trying out something that can potentially uncover a data corruption issue in their live system.
More likely, they paid someone to do blackbox security testing on the live site. In that case, they would be pretty lucky if all someone could find, was a way to digg stories multiple times.
I'm with you --- Digg's response is pure bullshit. They have $30MM in funding and don't have a test environment??? Yeah. Right.
Very true. If it was really testing, why would they stop exactly and immediately when informed by the blogger that their behavior was 'caught'? They would've continued the so called experiment if it was really one.
I think they did it to inflate the number of diggs on the front page stories so that it looked like more people were frequenting the site. That's the reason that there's no apparent bias in the selection of media outlets whose articles the accounts digged up.
>If it was really testing, why would they stop exactly and immediately when informed by the blogger
That's absolutely not evidence that it is not testing.
It's a good point, but it is in no way inconsistent with their explanation. It makes perfect sense because it was (theoretically) supposed to be a small test that nobody noticed. Once somebody noticed it, of course they're going to stop it.
It's consistent with other theories being floated around, but it's important to note that it in no way contradicts their statement; exactly the opposite in fact.
If your theory is correct, why not just do it more directly? Why not do it in a less obvious way? Why upvote random spam sites as well (some of which were created at the same time as the accounts in question...almost as if someone were testing submitting spam on Digg...)
I don't doubt it's possible, but it seems to me people are letting their bias against Digg get the better of their reasoning ability.
"Never attribute to malice that which is adequately explained by stupidity."
~ Hanlon's Razor
Well there's a couple problems here:
1) Let's say their excuse is bogus. So what where they doing? Gaming the system like ltgenpanda claims?
There's a contradiction there though. On the one hand you're saying that you don't buy it's testing because they're too smart and too capable to test on production...
...and on the other hand you believe they have to resort to this roundabout way of gaming their own system because they are not capable and smart enough to just do it a more direct way.
2) Your assumption (that they could do it in a test environment) is not necessarily true. If their explanation is to be believed, the testing is not the same as say normal software testing. The intent is to be a realistic security test...more like a "pen test"...which are quite often done on production machines.
So far I haven't seen anything that contradicts their explanation, and indeed it fits the facts as we know them pretty well, while the "game theory" doesn't.
Maybe some more conclusive info will come out.