I suppose it's possible that the company has changed so much since the senior engineers became senior that their ability to discuss the predictive power of VPN logs can't be relied upon. But it's also possible (IMO likely) that it hasn't changed all that much. So, to say that VPN logs can not have predictive power about productivity is false.

What is the alternative? The new CEO single-handedly interviews everyone, comes up with her own system of performance assessments, and makes decisions about thousands of engineers? That obviously can't scale. Maybe the new CEO could fire the entire management structure and replace it? Probably not scalable either, and you'd lose lots of good managers in the process. Just fire everyone? How can you not rely on people that already work there? That's just not realistic. You have to work with what you've got.

Keep in mind that everyone at Yahoo is not oblivious to the problems at Yahoo. It is possible for problems to be known, and to persist, simply because of company culture and inertia. Nobody wants to rock the boat. Nobody wants to be responsible for taking away a perk. But change the culture, maybe some middle management, and bring in a CEO that doesn't mind if people get upset a little bit, and these problems (that everyone knows about) can get fixed while still relying on people that were part of the old system.

First, North Korea is not off-limits to everyone else in the world. Generally tour groups have minders, so your discussions with the populace aren't exactly open and candid, but...

Second, tens of thousands of people have successfully made it out of North Korea in recent history. These people have all shared their experiences and have allowed us to learn quite a lot about how the country works and how the people think and what they believe.

Yes, thanks.

Generally senior engineers once did the things that the junior engineers are now doing. If I were doing this, I would specifically talk to senior engineers who once did the things that the junior engineers are now doing, just in case you're going to make the case that some senior engineers were hired at the senior level. This makes them distinctly (and possibly exclusively) experts on the question of "how much is VPN use predictive of engineer productivity"?

My speculation about how this could have been done rationally is simply speculation. If it's possible to do this rationally, that means the point that this must be irrational is false.

Are you trying to say that it is impossible to take a piece of data and apply it to a real-world situation? If it helps, I would define "carefully considered" to mean something other than "retrieve conclusion from ass".

Why do you believe they haven't detected it? Maybe they did, but couldn't make the tough decisions.

It's also not necessary to talk about productivity in absolute terms. Maybe the managers themselves have a good idea (possibly backed with real data) about the relative productivity differences between their employees. Maybe a trend toward poor performance from WFH employees was a trend that you could only see statistically once you started looking at the entire company. Possibly, any one team's performance disparity could still look normally distributed when looking at it locally.

I think that's up to Yahoo to define, yes? And presumably their more senior engineers know this? I guess it's possible that the respected senior engineers were brought in from the outside with no knowledge of how anyone else at the company works. If that's true, then my speculation about how they could have made this decision rationally would seem to be wrong, but that doesn't prove that it's impossible to make this decision rationally.

I think it is more likely that people at Yahoo know how a typical Yahoo engineer works.

Ask Yahoo.

Why is that necessary? Can you name a single company that does its employee performance ratings in a "verifiably objective manner"? I imagine Yahoo would rate engineer productivity in a manner that makes sense for Yahoo. If they can't do that, then how can they do regular performance reviews? It would seem like they'd have bigger problems.

My viewpoint is that your viewpoint--that the use of VPN logs for this purpose has no value--is wrong. It is not necessary for me to prove that Yahoo does things in any way that resembles my speculation. It is sufficient to demonstrate that they could do things that way. If VPN logs could be used in a way that results in a net gain for a company, then the assertion that VPN logs can never have value is proved false.

Are you under the impression that this is the only piece of information used to justify eliminating the WFH program?

According to the press. Do you believe what you've read in the press is a complete factual account of everything that transpired related to the WFH ban? Maybe she did say that. Maybe those were even her actual words. I rather suspect, though, that there was more to the conversation than that, and that questions were asked and answered that we don't see in the press. It seems improbable to me that the entirety of the Yahoo chain of management would accept such a thing with only VPN logs as evidence with no questions asked. It also seems improbable to me that Mayer (or any executive with an engineering background) would accept one metric like this and use it exclusively to justify a decision of this magnitude.

Your argument appears to hinge on the belief that you know at least as much as Mayer does.

Zero sense? Neither of us has the full picture here. You seem to be asserting that there can't possibly be a picture that allows this situation to make sense. I'm "desperately" trying to point out that there are solutions to this conundrum that don't require people to be utter morons.

It wouldn't surprise me at all if there were people that were productive working from home who are ticked off by the loss of their WFH privileges. Some of these people may even leave Yahoo, which means Yahoo's losing productive engineers. But it also seems quite probable to me that there are non-productive people abusing WFH. Some of these people don't actually want to go into the office and work, and so Yahoo is probably losing non-productive engineers as well. Of those that are left, returning to the office, some will be productive and others won't be. Even with the loss of productive engineers, it's entirely possible that Yahoo's productivity per person (as a rate) will go up. A net productivity per cost increase makes more than zero sense to me.

I think part of the problem here is that you seem to think VPN use is a measure of productivity. I think Mayer's point is simply that it's predictive.

The two aspects of my post are somewhat independent. For a single case where you literally have chosen the "best" option, there is nothing "better" for you to say they should have done if you have a bad outcome. My point is that demand for expensive services exists from two different directions: patients demanding something new and fancy (because they aren't paying for it), and doctors demanding unnecessary and conservative (because they pay if they miss something). Another way to look at it:

1. New technology gets invented
2. Doctors don't use it yet because (a) it's not proven to be sufficiently effective or (b) it is, but it's not worth the expense
3. Patients see "ooh shiny" and take their business someplace that provides it, regardless of 3(a); or
4. Patients learn about it and say "that could have saved my loved one" and sue because it wasn't used (despite 3(a) or 3(b))
5. Hospitals and doctors are now more likely to use it, either because this gets them more business, or because it reduces the number of people suing them for not

This is the key point. The American health care system is expensive because we demand expensive health care.

For those of us with insurance, we pick the best treatments, not the most economical. For those of us that can afford to choose what hospitals we get non-emergency treatment at, we pick the ones that have the experts, and the robot surgery facilities, and the fancy new MRI and PET scanners. Prices are set by contract with the insurance company, so why wouldn't we pick the one with the best marketing/facilities?

When we have bad outcomes, we sue the doctors, the hospital, the equipment manufacturers. We (via our lawyers) say things like, "they should have done more." This encourages them to practice medicine defensively: use the more expensive drugs, book more time on the expensive imaging devices, pay out settlements as a cost of doing business. And so, as time goes on, consumption of expensive health care rises as expensive health care options proliferate. In some ways this is good (sometimes the expensive options actually are better), but usually it's just wasteful.

It's easy to blame "free riders" and EMTLA, but this is a small fraction of healthcare expenses in the US.

I've never stated that; I'm just challenging your assertion that they aren't.

It seems more likely to me that they have:
- carefully considered how often a typical developer should need to use VPN, discussed it with respected senior devs and dev managers
- produced a hypothesis about correlating WFH productivity with VPN use
- ran the numbers (perhaps not just limited to VPN logs)
- pulled out a sample of cases that they predicted would be low performers for study
- confirmed their hypothesis
- considered the consequences of eliminating WFH (loss of upset high performers)
- made special considerations for those that deserved it
- eliminated WFH for the rest

All of that seems quite rational to me, and entirely consistent with what they've said publicly, and doesn't require anyone to be an idiot. Are you of the belief that all of the facts about this decision and the data leading up to it are the things that have been made public?

If someone asked me to try to measure WFH activity, and there were services like this that did not require VPN, I suspect I'd look at more than VPN logs. I also wouldn't conclude that low VPN activity means less productivity unless I understood factors such as these. Either Mayer and her staff are all idiots, or they're working with more information than they've chosen to publicly talk about. I rather suspect the latter is more likely true.

No; I code over SSH. I get it: there are some IDE-heavy workflows that make it more pleasant to code on a local workstation/laptop than remotely over a VPN. What I mean when I'm talking about company resources is stuff like:

- Documentation, design docs, specs
- APIs for other services I'm going to be interoperating with
- Code search ("Surely this is a solved problem somewhere else in the codebase")
- Distributed build systems
- Actually running my code in a production- or production-like setting

But, again, maybe Yahoo is just really different from what I'm used to (entirely likely), and so the things I can't imagine coding without are things Yahoo devs don't really use/need.

Interesting factual assertion there. Do you have a citation?

I think what you mean to say is, "I don't understand what useful information you can get from VPN logs on this. Hence the decision that I don't understand." It's easy to play armchair CEO from outside the company when you have no idea what data they ARE using and what other factors are contributing to their decisions. All you know is what they've chosen to publicly say.

Sorry, you're right. I agree that if they have specific data, they could use it in this fashion. And maybe they are, at least for individuals that are clearly "unsalvageable". But maybe they only see a productivity concern in aggregate? If the data show that, on average, WFH employees perform 10% less than WFO employees, but when you look at any one individual employee, it's hard to say there's really a performance problem per se, you could still make the case that moving everyone away from WFH is the right thing to do.

In theory, perfect management would be able to identify that 10% productivity problem in every employee. Even assuming Yahoo could eventually get to that point, it's going to take time to fix their broken management to do it. In the mean time, if you have data suggesting WFH (in aggregate) is holding you back, I think it makes sense to eliminate it, even though you're going to end up with good employees getting angry and leaving. So long as the ratio of good:bad employees isn't going down, it's probably acceptable. Big changes like this are always going to result in a shakeup.

When you reach the point where your individual contributors are being fairly assessed and management is reliably handling poorly-performing employees (and teams), it's possible a WFH program could then succeed. And maybe I misunderstood the article, but I don't get the impression Yahoo is there yet.

Why do you believe they haven't done that?

Why?

Why should it? Do you commonly expect companies to release all of the data used to support their decisions?

How does Google make "money out of the box" from ChromeOS?

Wow, my post got butchered by my use of less-than and greater-than signs. Let me try to replace the lost bits from memory: