Humiliation in Startups
mkrecny.com
mkrecny.com
If you're in charge, and your people screw up, it's not their fault. It's yours.
Yes, they made a mistake. But why did they make it? Did you not train them sufficiently? Did you not give them the mentoring they need? Did you fail to get important tools or information to them? Did you put them in a position of greater responsibility than they're currently equipped to handle, or where their skills don't match the needs of the position?
At first when you try to wrap your mind around this line of thinking, it can be difficult. Why should I be held responsible when Jack the Junior Coder is the one who actually screwed up? But eventually, if you stick with it, you learn something important: on a real team, who screwed up is irrelevant. If one person fails, the whole team fails. And your job as a leader is to help your people get to a place where they fail less and less every day, not to pin blame.
This is the difference between a manager and a leader: a leader is someone who accepts responsibility not just for his own actions, but for his team's as well.
I think startups have a lot to learn from the military community. Yes, for the most part modern militaries are extensive bureaucracies, but military thinkers and intellectuals are really at the forefront of management science. John Boyd basically wrote The Lean Startup 30 years before Eric Ries did.
The best general-audiences intro to Boyd-as-philosophy is probably Chet Richards' book Certain to Win: http://www.amazon.com/Certain-Win-Strategy-Applied-Business/...
Wikipedia has some brief descriptions of Boyd's work and thinking in his page on him: http://en.wikipedia.org/wiki/John_Boyd_%28military_strategis...
Fast Company magazine did a decent article on Boyd-for-business back in 2002: http://www.fastcompany.com/magazine/59/pilot.html
Many of Boyd's original writings have been preserved and made available for download here: http://dnipogo.org/john-r-boyd/ (Warning, though, these were not written for a general audience and can be quite dense.)
If they're unready, why did you put them in a position that required someone who would hit the ground ready?
It's also worth noting that the original Mac (the same time period most of those 'crazy Steve' stories originate from) was a huge commercial 'meh'.
If you're interested in that story, Randall Stross' book Steve Jobs and the NeXT Big Thing (http://www.amazon.com/Steve-Jobs-Next-Big-Thing/dp/068912135...) is a good telling of it. (It was written before Jobs' return to Apple and subsequent huge success, so it's a fascinating artifact of the time before he became Saint Steve.)
Also I didn't know (before that) that he refused to pay his daughter's tuition and some friends of his wife had to do it instead.
Awkward at the least...
This tendency to make Jobs into a person he was not it very disturbing. Of course, on HN you risk being banned for saying something bad about His Holiness because some of its highly ranked people are also members of the cult of Apple, but that alone should not be a reason to tip-toe around the obvious.
Ya, it's a stupid bug, but the way the manager dealt with it was totally out of line. Whether or not that "style" of management results in better output is moot, you just don't treat people like that. It's not like the developer shot the manager's dog or anything.
For example: Don't criticize, condemn, or complain. and Let the other person save face.
That is, criticize in private, give praise in public (give the person a fine reputation to live up to). Calling someone a retard is a counter-productive thing to do.
0: https://en.wikipedia.org/wiki/How_to_Win_Friends_and_Influen...
1: http://news.ycombinator.com/item?id=5584 (can't find where I read it, but this thread points to pg recommending it)
I think there's a lesson to be taken here. Not for managers (clearly good management would criticize privately) but for the employee. Learn to take a yell. It's not good style but it's widely common, people above you scream at you when you screw up. Learn not to take it personal, and not to let such incidents affect your passion for your work. The yelling is bad form, but it's not direct criticism at you. In large entreprises, it's the result of a chain of screaming starting at the top. In startups it's the result of the nerve-wrecking stress, founders go through.
'Constructive criticism' is an advice we give would-be managers, but it's a luxury starting employees rarely get. You're lucky if you meet one such boss in your career. Learn to take criticism, the bad kind, leaving out the humiliation. I found the best way to cut a yelling short is by answering loudly: "You're right I screwed up, I'm sorry. At least now I learned it, it won't happen again". After getting yelled at like this a few times, you'll realize it's really not against you, and such an answer would calm down your boss quickly; make sure not to make that mistake again, because then you'd really deserve a humiliation ;). This is easier said (written?) than done, but it's a natural part of one's career. I take it as a mental strength test.
Also, you won't be the junior on the team forever. One day you'll be responsible for an intern screwing up. Remember that moment and don't be too harsh.
Saying something that weird will just make it seem like you don't care or you're weird.
The fact you say you 'find it best' suggests you get shouted at a lot or you are making this up. Either way, you shouldn't be dispensing this kind of nonsense advice to other people who may well follow it.
You need to grow up.
> mistake, the initialization of variable $user (arrrgh!!).
> The website crashed on all the signed-in users.
I'm trying to imagine a professional, well run software shop that would actually hold you responsible for this. I can not. It seems to me that for this to have happened, many things must have gone wrong, or many professional practices must have been completely missing:
1.) there were no code reviews
2.) there were no unit tests
3.) there were no functional tests
4.) no one noticed any problems on the development site
5.) there was no QA person, or the whole QA team somehow failed to test this
6.) there were no 3rd party testing services that might have caught this (like Airbrake), nor any in house error logging services (errbit)
I could go on. I'm trying to imagine how a programmer comments out the $user variable and that change makes it all the way to production without anyone noticing. I think a lot of things have to wrong, all through the organization, before that becomes possible.
I read your story and I have 2 quick thoughts:
1.) I have made many worse mistakes (thought I was changing database password for dev, in the config file, but actually changed it for production)
2.) None of my worst mistakes ever made it to production, because the company had some process that caught them before such a change ever got pushed to production.
No programmer can avoid mistakes. That simply isn't possible. To insist that a programmer always write perfect code would be like insisting that an NBA basketball player make 100% of his 3 point shots: the world would be a very different place if it was possible to insist on perfection.
A well run organization has multiple checks that ensure mistakes made in development don't make it to production. Any organization that puts 100% responsibility on one programmer is taking a 100% certain risk that eventually some terrible mistake will make it to production.
Some organizations put more responsibility on individual programmers and some organizations put less. That is fine. A range of strategies can be defended. But if an organization wanted to shift most of the responsibility onto a single programmer, then they would also have to give that programmer broad latitude regarding time, so that the programmer can implement whatever they feel is necessary to ensure the robustness of the code (for instance, unit tests, functional tests, setting up automated uptime checks, etc). I'm guessing, from what few details are written above, that the above commenter both had responsibility for the code but was also under pressure to move quickly and shove stuff out the door. Under such circumstances, management needs to take 100% responsibility if something goes wrong.
Now I will qualify everything I said above with an anecdote: In 2005, when working at monkeyclaus.org, we had a little startup where I was the only programmer and we would push my changes straight to the "live" site. But we hadn't really launched yet and the few dozen users we had were just friends of ours, so the "live" site was really a "dev" site.
I had extensive experience with bullies in school and let me tell you being submissive or showing fear of any kind only brings out the worst in them. I know you can't throw the boss against the wall (which would get a school bully of your back) but at least don't make it worse.
And no, I hope you will not stay long enough in a place like that. There are better jobs elsewhere. Shoveling shit isn't as bad as that.
Of course if it was a one time thing and the manager apologized (in public) that is another thing.
On the other hand, guys I really respect just call those "rookie mistakes" and expect them to happen. Hopefully you catch them in code review before customers are impacted.
This is written in law, and has been supported by court cases.
One incident like that is wrong, but could perhaps be explained by someone having a really bad day. Recognition, apology, and no further incidents would help.
But continued incidents? The company is leaving themselves open to lawsuits. ("Constructive dismissal" and employment tribunal in the UK.) I'm pretty sure that toxic work environment leading to poor health is something that has been through US courts.
I mention this because sometimes the only way you get through to PHBs is to talk about costs and risks, and not "don't let employees be dicks to other employees".
I'm not agreeing in any way with the manager in this situation; public humiliation of employees by managers is both morally cowardly (bullying someone from a position of power) and ineffective over the long run (in my opinion). But still, stress happens in most jobs. Often times it is caused by a combination of home and work factors. It's ridiculous to make "preventing stress" a legal duty of employers, because that's not wholly within an employers control. Example: Faulty machine used to serve customers blows up at work at now you have a lot of angry customers. Stress ensues.
Employers are expected to have procedures in place to deal with things like sexual harassment. This is similar, they're expected to have procedures in place to deal with bullying.
I agree that I might have made it sound stronger than it is. In real life it's pretty reasonable.
Making sure employees are happy and safe in their work is good business.
The people on that page that are saying the abuse as well deserved are part of the problem.
Sure, it's not good code, and causes problems in production. It's certainly not the right way to do this -- but constructive criticism, gentle corrections, and rewards/acknowledgements of successes when they occur are far more effective in the long run than this public tirade bullshit. AFAIC, the PM that humiliates an employee in public for any reason is (a) immature, (b) short-sighted, (c) unprofessional, (d) a poor choice for a PM, and likely (e) insecure and feels the need to strut his/her intellectual "superiority" in front of others lest their authority/superiority be called into question.
I also would have quit on the spot -- life is far, far too short to deal with that kind of crap, and there are other work environments in our field where one can do what one enjoys and where it's okay to be a fucking human being instead of being viewed as merely a code generation resource that can be kicked around when it doesn't "behave."
The proper response to that kind of public humiliation is "Okay, you're right, it was a pretty stupid thing to do. But not as stupid as abusing your employees and creating an environment where the primary motivators are grounded in FUD. I'll collect my things from my desk because...how to communicate this part adequately...oh yeah: fuck you."
Is this a source of concern for the DB? If it is, you might want to consider limiting connections on one side or the other at a deeper level. Self-inflicted DoS bugs during development aren't fun, but actual DoS attacks in production are even less fun. If there's a potential way to bring your DB to its knees and you get popular enough for trolls (let alone actual crackers) to take notice (and if you're writing video games that presumably have high score boards or something similar this isn't that unlikely) expect people to find it.
It's good to get yelled at some number of times throughout your life to help you grow a thicker skin, but I've never been yelled at in the work environment. If it occurred I'd be strongly motivated to quit like others here. Stating matter-of-factly "That was stupid" is one thing, joking and exaggerating stupidity of yourself or others is another (and is dependent on implicit understanding of such and also requires a certain culture of the group), but when it gets to a manager actually screaming at you about a possibility of something bad happening then that's the line, at least for me. If I wanted to get screamed at, I'd join the military.
Anyway, I was interested to see what was happening with that page out of interest so I checked out the source.
What I found made me laugh so hard. Given the fact that I knew that second perfect cache invalidation is pretty hard...and given time zones... I knew that this wouldn't end well. What happened as a result? They created their own distributed denial of service that was set to fire off every connected user at the same time. They ended up taking down their new launching page for their phone.
This was the code:
if (timediff < 0) {
//console.log("LIVE!");
location.reload(true);
} else {
// Do stuff.
}
I wonder what kind of jokes they had at Samsung over this one.
Code reviews, instruction, and collaboration help make better code.
And in cases like this ensures further humiliation.
Instruction can work, but it requires a smart programmer who can do it right. Collaboration can work but it reduces the speed you can think at to how fast two people can communicate instead of how fast one person can think.
And no, I don't have good solution (other than to ensure you only hire the best people).
Anywhere from 16px to 22px would be fine, we are here to read the text afterall. It's the focus!
(ctrl + mouse-wheel for me)
e: VVV Good point.
> so afraid of being publicly railed-on that they wrote pretty much bug-free code all the time.
That's a non sequitur. Programmers try to write bug-free code all the time anyway. No one tries to introduce bugs.
This is an interesting statement in the article and I think it is worth bringing up. It contrasts significantly to the other article I just read that said that "public shaming", which is just another word for public humiliation, is a desirable practice needed to stem the existential threat to engineering of "brogrammers".
http://www.cnn.com/2012/05/10/opinion/trapani-brogrammer-cul...
> Sometimes the road to enlightenment is paved with public shaming.
Tags like "stupid," "bad at ____", "sloppy," and so on, are ways of saying "You're performing badly and I don't know why." Once you move it to "you're performing badly because you have the wrong fingerings," ... it's no longer a vague personal failing but a causal necessity. Anyone who never understood limits will flunk calculus. It's not you, it's the bug.
Wow, was it that hard to keep it down a little? see:
"Dude there's no delay in your while loop. How many times do you think the client can connect to the DB per second? Thousands! go fix that"
See? is not sugar-coating it, simply not being a gigantic asshole about it.
I too have been the subject of manager outbursts like this. Only do it if you're deliberately trying to get rid of the person, and if you're doing that, be aware of the precedent you're setting. There's too many things to know about programming for any one person to be expert in everything, and never make a WTF mistake.
What will you do differently next time?
Would you call this behaviour out to your other manager or directly to the perpetrator?
If you want to avoid it happening again you should speak up. Unless you don't mind the abuse?
"This is retarded! There's no delay in the while loop. How many times does the client have to connect to the DB per second? Thousands! I can't believe this implementation is so fucking stupid!"
In this case we try to refer to the code directly, taking our reference to the person.
If you think public humility is motivating it probably is in the short term. But its the most motivating for getting that resume spruced up again.
It's crazy to be mad at a programmer for a bug in their code found during a code review. Discuss it, learn from it, and not only does the code get fixed, everyone involved becomes a better programmer by some small amount.
The only time it's justifiable being mad is if something gets through code reviews and into production - and at that point it's not the programmer's fault, it's the fault of procedure which enabled buggy code to ship undetected. So you get mad at the system, review the failure and fix it.