Hacker News Reader

Comment by cromwellian | Hacker News Reader

You can't iterate on products without launching them. Startup culture in Silicon Valley is all about failing fast.

The heart of science is running experiments, and if Google can't run experiments or pilots, it will fail to the innovators dilemma. Our democracy is similarly unable to run experiments to test out policy, which was another thing Larry talked about, as a result, we argue over the effects of policy without doing anything. States were supposed to be the laboratories of democracy, but even on the state level, hardly any experiments are conducted.

The idea that you should only launch services you plan to run forever is a plan for stagnation and death.

View on HN

Yes, but michaelochurch was talking about something else.

Many of the Google products you see today were 20% projects. Google had no idea GMail would be a hit. You can't simultaneously criticize management for 20% time being more difficult, and then criticize them for launching lots and lots of experimental products and then killing off the ones that don't work.

We agree. I am all for "lots and lots of experimental products".

Calibration scores (secret performance reviews that happen in the Perf Room-- yes, Google has a physical place for "calibration" that is actually called "The Perf Room") are not experimental technology or product. They're an experiment (and a failed one) on people. Different rules together.

Odd. I've been doing eng management at Google for the better part of a decade and I've never heard the term "Perf Room" before, nor did an internal search turn anything up. Maybe it is used, but at the very least I'm unaware of it.

I realize you worked there for a few months a few years ago, but I think someone might have been pulling your leg if they told you that phrase. (But you're right, like most meetings at Google, calibration meetings do happen in rooms. They're just normal conference rooms, though. Nothing special.)

For everyone else: calibration is a process where other managers at Google cross-check the performance scores given to employees across different teams and parts of the company. We do this so that an individual manager can't introduce undue bias or play favorites, and so an engineer at a given level in one part of the company is more or less the equal of an engineer at the same level in another part.

Calibration is one part of the system I think works really well, much better than any other place I've worked. In fact, I think more companies might want to emulate it.

Why can't Google just grow a pair and go to full-on open allocation?

If people actually get together in conference rooms to conspire against their employees and wreck peoples' careers, as you've admitted they do, that's not only wrong but a sign that lots of time is being wasted.

> Why can't Google just grow a pair and go to full-on open allocation?

From the description, calibration seems to address the issue of consistent assessment across the company. Open allocation, whatever else it might have going for it, doesn't seem to do that (and for open allocation to work effectively in a large organization, consistent assessment would seem to be, if not essential, highly desirable; even without headcount concerns, churn between projects and getting new people up to speed is a cost that needs to be justified by expected value.)

> If people actually get together in conference rooms to conspire against their employees and wreck peoples' careers, as you've admitted they do

That's actually not what the poster above you said, and I really think that if you can't read anything about Google without being distracted from what is actually said by the white-hot heat of your pre-existing hatred (however justified that hatred moght be), you probably should just avoid participating in any discussions about Google, because you aren't going to be able to contribute productively.

Does GMail make money? When it first came out, it seemed almost like a gift to the community which they made because they needed email and built it. It was great for email, but it was hard to see where the piles of money would come from. Now, I think they've become more business oriented and try to extract revenue from everything.

Reply on news.ycombinator.com