Hacker News Reader

Comment by patio11 | Hacker News Reader

Sadly, Goooglers are not unaware of this issue. That's the trouble of problems with Google: at J. Random Megacorp this would be a stupid oversight. Multiple Googlers have brought up internally that this policy would result in them locking accounts of innocent users and they lost the argument. You're quite literally acceptable collateral damage of a core design goal of Google+: for strategic reasons, they want real identity relationships not "Internet identities" which may or may not be pseudonyms. They want Facebook names, not IRC nicks, and they're willing to backstop that with their famous willingness to use individualized customer service. These decisions were internally controversial but supported at the highest levels of the company.

View on HN

Oh, I realize that. I even agree with their reasoning.

I just don't like the fact that their algorithm is so bad that I came up as a false positive. Algorithms is the one place you don't expect Google to fail.

I mean, come on, the have thousands of my emails, my geolocation, my facebook profile is public, and they can't figure out Or is my real name?

They haven't failed. They have a huge body of identified individuals that they can sell to advertisers, efficiently and economically harvested.

Oh, they failed at individual user service? What farmer is concerned about an individual cabbage? It would cost way to much to get down off the truck and carefully retrieve the few units that rolled off the truck.

There is absolutely no failure here. None.

I believe what the OP is saying here is "failure" occurred not on a strategic corporate level, but on a computer-science-algorithm-geek level. He hasn't disclose his full name, so we cannot fully judge, but how would you react if "Joe Smith" came out as a false positive?

Of all the crazy things we're hearing about Google lately, failure on an algorithmic level may be the last thing we expect.

Then again, maybe the name is indeed more subtle than the OP seems to think...

And I'm saying that the algorithm is imperfect by design, and accepted as such, because perfection would cost too much in computing and human resources.

You cannot fail a cabbage.

I'm sure cabbage truck designers pay a lot of attention to the cabbage loss rate... Look how much effort was put into the design of shipping containers.

> What farmer is concerned about an individual cabbage?

What a great analogy! We don't think twice about the farmer leaving a few cabbages out in the field, but we get all up in arms when we realize that we are that cabbage.

OMG, a talking cabbage!

> Algorithms is the one place you don't expect Google to fail

Maybe this thing (the reality of names) is not rationalisable, and therefore not modelizable with an algorithm. For instance, I know someone who has chosen the name of a Russian president has his nickname, and I guarantee you it is not his real name. How would Google detect that? On the opposite, "Ng" is a perfectly valid surname in Southern Chinese, and it really looks like a nick...

This is not true. Unfortunately, I can't elaborate further.

Ah, the classic "If you knew what we knew, you'd be okay with torturing terrorism suspects" argument popularized by multiple US administrations. A blanket assertion that due to your secret knowledge of the situation you can declare that all the statements by ex-Googlers about the internal debate and the references by Google executives to careful consideration of the policy are false is on the face of it unproductive. What conceivable secret reason for its name policies could Google have that couldn't be revealed to the public without damage to the company, but would be convincing if it was?

I understand that you have internal knowledge of the debate within Google and the reasoning for the policy, but it doesn't contribue to the public discussion to make an unsupported statement without evidence or or argument beyond the unstated appeal to name recognition. The information content beyond "trust me" is nil. Please, you've proven you can do better than this - if you can't speak about the issue, don't speak about it.

Reply on news.ycombinator.com