When the services go down, you can't log in to the relying sites. Luckily, core infrastructure like the account systems is a very high priority for the engineers, and the big providers have plenty of resources to keep them up -- and they do. My bank's site is down far, far more often than Google's auth servers, for example. How much more often? I don't know... I've never seen Google's auth servers down.

Pick the top several and you'll cover nearly everyone. For the tiny percentage of users that remains, you have to either offer password auth (which means all of the work and risks of maintaining a password system, but at least when you screw it up only a tiny percentage of your users will be affected) or push them to get an account with one of the providers you support.

You're promoting perpetuating a long-standing, widespread and hugely-damaging user security error in order to avoid a relatively obscure problem which can actually be fixed through purely technical means. Not a win.

What you describe as a problem is actually part of the solution. The problem with classic OpenID was that it was virtually impossible to get, say 1st Bank of MyButt, to use it, because absolutely anyone could be an identity provider. I personally agree with you that classic OpenID was better in that respect, but 1st Bank of MyButt doesn't. They're hemming and hawing about letting Google manage their user's identities, but they will at least consider it.

Heh. Your snark falls rather flat, given how egregiously wrong the parent was.

Putting them in password managers is the right thing to do.

You're wrong. Hilariously so.

The entropy of "thatswhatshesaid" is far lower than 43,608,742,899,428,874,059,776. Randall Munroe calculated correctly in the XKCD comic, of course. He didn't assume that each letter was random, he assumed he was choosing four words at random from a dictionary of a specific size (about 2048 entries == ~11 bits of entropy per word). Your password is clearly not a selection of randomly-chosen words, and even if it were, it would likely have been from a much smaller dictionary.

This highlights the danger of asking users to pick passwords... even those who think they know what they're doing are likely to screw it up. Munroe's advice in 936 was good... but I think it has mislead more people than it has enlightened.

No, it's much better to use a password manager and let a computer pick large random passwords for you.

Parent was modded funny, but this is what your passwords should look like -- long and random, and typing them is a PITA. Any web site that disables pasting or prevents your browser or extensions from auto-filling passwords is broken. The sad thing is that most sites that do this (other than those that do it by accident because the devs are clueless) do it because they think they're increasing the security of their users' accounts. They're not.

Solutions like LastPass et al are the best, but honestly just using your browser's password database is better than reusing passwords everywhere. And Chrome and Firefox (at least, perhaps others) offer the option of keeping your passwords synced to all of the devices you use, optionally protected with a master password. Browsers need to offer password generation as well. I think some are working on it.

Of course, the real solution is to get rid of passwords. Web sites should switch to using OpenID authentication. Yes this means that most users will use their Facebook or Google logins, which means that, essentially, the site has outsourced its account security to those other entities. So what? If the developers of random web sites think they can do a better job of account security than Google or Facebook -- they're wrong . I work for Google and previously spent a decade as a security consultant in the financial industry and after seeing how they all work from the inside, I would feel much more secure about my bank account if I could use my Google account (with 2FA, plus all of the analytics and monitoring Google does) to log into it rather than trusting the bank to do a decent job with password-based security. I haven't seen Facebook's infrastructure, but I know people who work there, and they're good. Far better than you'll find at a typical bank, much less J. Random Web Developer.

FWIW, I'm not a member of the community. I play Wesnoth off and on for a few weeks every couple of years. I also like the random element and don't much care about characterization, world-building, lore or storytelling. Not that I don't like those things, just that Wesnoth is more of an occasional light diversion for me, so those things don't mean much.

Sure, it does, which is why managers convert such things into spot bonuses -- which are generally several thousand dollars.

The downside of rewarding primarily with peer bonuses is that it might create a culture of doing good stuff for peers in order to collect peer bonuses rather than doing good stuff for peers because it's abstractly good to do good stuff for peers. I don't know how real it was or was not, but I have heard that some obnoxious Googlers with special skills or access or knowledge made a habit of demanding multiple peer bonuses before being willing to do some task for some other team that needed it. The "one bonus per" rule pretty much eliminates that because -- to people as well-compensated as Googlers -- a single $125 bonus isn't worth the overhead of negotiating; it's much more effective to just be "nice" and do stuff for people who need it, gathering the occasional peer bonus and lots of kudos, as well as building the network of people who will offer support at promotion time and/or help you out when you need it.

The effect is the same: it incents employees with special skills or access or knowledge to help their peers, but makes it more of a "gift economy" where everyone tries to be helpful to others in expectation of eventual good karma coming back to them, rather than one of barter and bargaining in which people jealously protect their advantages.

"One front-line manager" != "management". The latter implies higher levels of company leadership.

Spot bonuses are generally much larger than $125. The spot bonuses I've received have been several thousand dollars each. Peer bonuses are $125. And Googlers don't pay for meals :-)

(I do pay for my meals, but that's because I work remotely. So I don't get all the on-campus perks... but I also don't have the insane cost of living.)

It's orthogonal to that narrative. It could be that Erica's manager decided not to give her the bonuses because she's a woman, or because she's black, and the other manager decided to give her colleague the bonuses because he's a man, or white (assuming he is -- I don't know, but it's probable given the demographics). Or not. Given the vast array of possible reasons for the two managers to choose differently, I don't see any reason to assume it's because they're bigots. My wild guess is that her manager was annoyed by the spreadsheet and the other guy's manager thought it was cool, so her manager found reasons in the rules to reject her bonuses until beaten down by the volume, and the other manager approved them all.

Seems far more likely than sexism or racism to me. But I could be wrong. We don't know, and never will.

Management didn't "freak". The spreadsheet in question is alive and well, and Google employees continue adding their information to it (I did). If management really wanted gone, it would be taken down. Erica Baker's manager wasn't happy about it, and she was invited to talk to her manager about it. It may or may not have bothered someone above her manager; Erica doesn't know and neither do we.

Her manager also chose to interpret the peer bonus rules such that the bonuses peers sent her forward weren't given to her. That's at least partly correct on her manager's part. The peer bonus rules say that any given action/effort can only be rewarded once. If the manager feels that it was a really valuable contribution the manager can choose to discard the peer bonus ($125) and instead award a larger spot bonus (amount variable), but only one peer bonus per act.

What is a little bit weird was that Erica said peer bonuses were rejected before one was approved, so the rejections before the approval weren't due to the one PB per action rule. Also weird is that Erica said her colleague got multiple bonuses for the spreadsheet. That shouldn't normally happen.

You're presumptuously assuming that I don't understand that there are exceptions.

The approach I recommend will, however, work for the vast majority of employees, assuming the necessary apps exist or can be built (or front-ended... ick, but it sometimes is the best option). Then, with the majority use cases out of they way, the security team can turn their attention to dealing with the special cases -- isolating them, locking them down to the degree possible and monitoring what can't be locked down. Or, in really special cases, training the users and making them responsible for their own security. That last tends to be the best option with developers.