Are Overemployed 'Ghost Engineers' Making Six Figures to Do Nothing? | Hacker News Reader

Are Overemployed 'Ghost Engineers' Making Six Figures to Do Nothing?

404media.co

13 pointsPlasmoid26 comments

> The Stanford research has not yet been published in any form outside of a few graphs Denisov-Blanch shared on Twitter. It has not been peer reviewed.

While I don’t doubt the existence of some useless engineers and people secretly working two engineering positions, the fact that they’re claiming 9.5% of engineers are useless based on the output of some code analysis tool and the paper isn’t even published yet makes me think this is junk science. Is their tool going to mark the engineer with decades of institutional knowledge who mainly spends their time helping others as useless? Or the tech lead who is a manager in all but name but doesn’t have time anymore for code?

Allegedly their algorithm controls for that, and companies have explicitly given them a list of employees who can't be measured by commits alone:

> I asked Denisov-Blanch if he thought his algorithm was scooping up people whose work contributions might not be able to be judged by code commits and code analysis alone. He said that he believes the algorithm has controlled for that, and that companies have told him specific workers who should be excluded from analysis because their job responsibilities extend beyond just pushing code.

Companies will definitely do an accurate and exhaustive job of that.

No company above trivial size would be able to produce such a list without a significant amount of work (you'd need to survey every single EM, at the very least), so I would, ah, have doubts.

Someone commented in a prior thread that they spent an incredible amount of time tracking down a bug that was resolved in two characters of hex. Surely, that person would be marked as a lazy engineer who should get fired. This is meaningless.

I like the way it's phrased in an old quote:

"If I had more time, I would have written a shorter letter."

"... Moreover, I am reluctant to break this already sealed envelope, as you undoubtedly now have done, just to replace the letter with a shorter one."

> When I asked Denisov-Blanch if his algorithm was being used by any companies in Silicon Valley to help inform layoffs, he said: “I can’t specifically comment on whether we were or were not involved in layoffs [at any company] because we’re under strict privacy agreements.”

> I have friends, I call them at 11 am on a Wednesday and they’re sleeping, literally. I’m like, ‘Whoa, don’t you work in big tech?’ But nobody checks, and they’ve been doing that for years.”

> [he] said that he imagines a future where engineers are judged more like salespeople, who get commission or laid off based on performance.

It's amazing how much this guy clearly has an axe to grind about his "friends" and is apparently rewarded for tweeting out unsupported conclusions like "might work multiple jobs" to unpublished, un-reviewed research.

Do they not do IRB review at Stanford? Why are you allowed to enroll companies in research by touting that they can use your algorithm to lay employees off?

*Technically not a layoff if it's based on the employee's performance metric but this distinction seems not to matter to the author

My offer still stands: I will do nothing for half the price.

I’ll do even less for 4 x the price. Does this make me a consultant now? Oh sorry I forgot to mention AI

Yes, but are you also a thought leader?

I’ve thought about it. Posted about it. I can safely say I’m a self appointed and self identifying thought leader.

It sounds like this measures "code monkey" productivity, right? Which is fine if you're only running it against your code monkeys, I guess - but how many of those can you possibly have?

Should be fairly obvious to any competent manager, or team member, who is contributing regardless of lines of code and who is simply doing nothing. Don’t need a fancy analysis tool for that.

If they want to reduce head count, then reduce head count.

If they want counter-productive results, then anoint this Magic Algorithm to do the personnel management, and watch the (purportedly) 90% of productive engineers wast a bunch of time trying to game the Magic Algorithm, if they feel it needful.

If the managers are so checked-out that they can't even equitably distribute the work their teams have to do, it's really hard to see how any software no matter how magic could help. Seems like it would just give them excuses to be even more checked-out.

"9.5% of managers are useless; might work multiple jobs."

You mean 99.5%, right?

dev jobs I could probably do two easy - manager jobs I can probably do eight, 1/hr per job is plenty

It’s interesting the study and toolkit is focusing on engineers. What about engineers management?

Has this Stanford academic ever worked on a production system?

If the only tool you have is a pen, the whole world looks like publish or perish.

My guess is that he will soon depart academia and announce his new workplace surveillance startup.

As you get promoted in the big companies you're expected to take more of a mentorship role, doing design reviews, leading teams, running and attending meetings, writing documents, working out conflicts between teams, designing features, designing architecture, triaging bugs, reviewing code, etc., etc.

The more you rank up the less actual coding you're expected to do. At my last job, after getting promoted to senior engineer I was at times going for multiple month periods without writing any actual code.

I'm not saying it's good or bad, I'm just saying there's a lot of senior software engineers that don't do much actual coding. The tool might have recorded me as doing nothing when I was all day working on non-coding stuff.