If that is (and had historically been) true, what is the article about then, or do you believe it's inaccurate? The whole thesis is that people are now being asked to be productive who were not previously held to productivity

Edit: or do you mean that previously there was room for slackers as long as they didn't care about promotion, and the difference now is their jobs are at risk?

Google is an incredibly complicated situation. Yes, it's impact-driven. But many other variables affect how your impact is perceived. For amusing reasons, over the past decade, the number of people who can and want to work on the core parts of google3- the load balancer/front end/web server/search/ads- has been dropping. No CS grad wants to work on a 15 year old web server that is mostly patches to deal with odd cases and a million tests that fail if you look at a load bearing comment oddly. The dwindling pool of experts has seized the memes of production (ha) and has the ability to work less because they are absolutely needed.

The best anecdote I ever heard from an SRE was "the VP was chagrined when I told them it didn't matter I was in a nerf war, because nothing was broken or needed to be fixed, but they still saw my point and let me continue" (I used to work in an SRE area that overlapped with an SVP's preferred meeting room so Susan Wojcicki was always getting pinged by a rogue nerf bullet).

As for slackers? Sure, when I joined around 2008 there were definitely folks who were slacking off. in fact they got rid of sabbatical because so many people were taking it then coming back and quitting.

Going to have to steal “load bearing comment”

Nothing wrong with slacking off unless you have a real job as a firefighter or ER nurse.

Even firefighters also sometimes slack off when they're not responding to a call, although the responsible ones do it in a way that allows them to go from slacking off to driving off in the fire truck in a ridiculously small number of minutes (like 2). As you say, nothing wrong with that.

I once walked through a park on a fine spring day and saw an EMT team playing whiffleball using their stretcher as a backstop. Great way to wait for a call if you ask me.

Considering their low pay and high responsibility, I would never take away what little they have such as a little park time.

> or do you mean that previously there was room for slackers as long as they didn't care about promotion, and the difference now is their jobs are at risk?

Yes, I think they collected lots of mediocre employees through the aggressive hiring.

Also, some of them may be not productive by common metrics (say bugs closed, lines of code implemented), others are good at those metrics, but produce high quantity/low quality work, and make more damage than impact: systems are overengineered, buggy, etc. I think second type is likely makes companies suffer much more than first one, they are much harder to trace and fire.

> Google also has big impact focused culture, if employee doesn't produce business relevant impact, he is not rewarded (bonuses, promotion).

Xoogler here.

I see how people may view Google as having an impact-focused culture, but I would argue that Google has a perception-of-impact focused culture.

Google historically has sought to hire clever people-- clever people that understand and care about optimization. It should be no surprise that Googlers also applied their cleverness to optimize their behavior toward what they understood about the reward system called perf/promo (some of those understandings weren't necessarily correct, but when enough people share an understanding or behavior, correct or not, it can have a reflexive effect on the system itself).

Google has reached a critical mass point where values it cultivated in its people through these reward systems now drive the culture and common behavior.

There are a few major components to this:

Perceived Impact: Perf/promo-optimizing Googlers optimized for things that they understood or believed to be perceived by promotion committees as "impactful." This does not mean that those things are actually impactful, but perceived as impactful.

Impact isn't Technical Skill: What counts as "impact" and the need for impact varies by an employee's ladder (job family) and their level. Generally for individual contributor SWEs, levels were meant to be representative of an individual's technical skill/ability and leadership. Generally, assessing and gating a technical individual contributor's performance rating and promotion based on business impact doesn't lead to fair or correct outcomes for the person when the business impact isn't associated or driven by that individual's work. Also, from the reviewer's side, a person's work/contributions generally needs to be viewable as the driving factor for the impact portrayed. This creates a bias toward things that can be seen as 'technically complex' or 'technically challenging.'

Perceivability and Overemphasis on Visibility/Measurability: Many important things in life and engineering are not easily measurable, nor can the explanation of how to appreciate the "why this matters" be condensed into ~300 words. You have likely had scenarios in life where your intuition, experience, or judgment for what the right or good thing to do made something obvious for you, but the explanation would probably take too many words to fit into a promotion packet or a few minutes during an short interview, so you'd have to select obviously understandable and relatable examples when sharing with others -- This drives a bias toward doing things that can be easily perceived and visible to group of reviewers in a calibration and promotion committee.

Lowest Common Denominators: "Perf" (performance review) ratings and Promotion are separate processes. Performance ratings are recommended by managers, and calibrated/adjudicated amongst a set of managers from the same calibration group, typically people from your own team or adjacent teams. Promotion nominations are adjudicated by small committees of ladder (job family) representatives drawn from across the company (this was later changed to representatives across the person's Product Area/Division as the company got bigger). Sometimes the important things for a person to do are contextually related to the dynamics, circumstances, or needs of the local organization, but those very things wouldn't be appreciated or understood by a person in a similar role elsewhere (see previous paragraph), so a person often has to make a tradeoff on where to apply their time -- doing the right thing for the team/organization, or doing something that could be more generally perceived as impactful by a committee of people that don't know you, probably wouldn't appreciate why fixing ___/paying off technical debt/etc was important, and might not understand that the nature of your work is very different where you are than in other teams/orgs/product areas.

Overburdened and Sometimes Lazy Reviewers: Calibration group participants aren't volunteers, but promotion committee members are. Both groups often have more packets to evaluate and assess than they have time to commit while also doing their regular jobs. In "perf" and the promotion process, the person being reviewed typically has to author their own statements for why their work represents the characteristics of strong performance for their level and/or the next level -- those statements can often be a lot, or hard, to read. Checking and verifying that a launch happened and that a person had associated code (CLs) and design docs is a lot quicker and easier than reading through a bunch of code, design docs, etc and trying to verify the "impact."

So you could imagine why "I launched _______" becomes a common focus for perf/promotion optimizers, and a common understanding amongst the larger community on what's important to be rewarded...

What I find enlightening about your description is how closely it matches incentives and resulting behavior in much less technically complex or profitable professions. Every dynamic mentioned is basically the common sense understanding of corporate workplaces we gain by late 20's or early 30's, but with a few SWE idiosyncrasies added in.

In some ways, it's both terrifying and comforting that highly competent people in difficult disciplines are also like this.

Yes -- We have to remember that Googlers are people and do what people tend to do :)

So, to summarize: Google as a company tries to establish impact driven culture, but complexity of this task (hard to measure impact) and humans imperfection (laziness, politics) make this goal not easy to achieve.

Google as company tries to establish an "engineering-driven" culture where how things get done matter.

Years ago, they tried to temper the bias-toward launches with a need for actual "impact" -- this led to a common saying of "Landings, not launches."

We end up with an amalgamation of technical complexity and launches as demonstrations of impact instead.

The problem is... The visibility and measurement of "landings" usually takes time, and the amount of time needed can be well-beyond the horizon of when a regular-level employee needs to be able to have their performance and promotion assessed. The effects of some things may take a year or longer to see. And as said above, measuring actual impact isn't always easy.

Google has a big focus on technical excellence and complexity. Facebook is the one with the impact focused culture. You need to get your PRs approved by language experts at G, while at FB people frequently ship things that are terribly written and have few tests.

Anywhere where there are committees that decide who gets promoted are not impact focused.

I have bad news for you, then: Facebook, which you laud as being impact driven, does exactly this. They call them "calibrations" but I've been in them and I can assure you that they're committees with all that implies.

Oh they certainly still do have some bureaucracy, but you will never see an L3 at Google being promoted to L5 in 1 year whereas that is a (very rare) possibility at FB.

I specifically know a person who was double promoted from 3 to 5 in their second full cycle at Google. Just over one year.

Except that they over-reward "design a completely new project in market X" where they already have a product, and under-reward fixing the existing product in a backward-compatible way. That means customers can't count on Google supporting anything long-term.