Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Honey's Dieselgate: Detecting and tricking testers

vptdigital.com

Post reply on HN
Print view View original

41–50 of 175 posts

Re: Honey's Dieselgate: Detecting and tricking testers

#41

Post by petterroea »

Aurornis wrote: post #29
Waterluvian wrote: post #26

Earlier quoted context omitted.

> what the engineers thought when doing design reviews for a "selective stand down" feature. Possibly a version of, “I lack the freedom to operate with a moral code at work because I’m probably replaceable, the job market makes me anxious, my family’s well-being and healthcare are tied to having a job, and I don’t believe the government has my back.”

From my experience, it’s more likely that the engineers who got far enough in the company to be working on this code believed that their willingness to work on nefarious tasks that others might refuse or whistle-blow made them a trusted asset within the company. In industries like this there’s also a mindset of “Who cares, it’s all going to corporations anyway, why not send some of that money to the corporation that…

I suspect you are right. It reminds me of the whole "at the government you can hack legally" argument used by government intelligence agencies to recruit hackers.

I think a lot of skilled engineers want interesting challenges where they break boundaries, and being in an environment that wants you to break those boundaries allows them to legitimize why they are doing it. That is, "someone else is taking moral responsibility, so I can do my technical challenge in peace"

Re: Honey's Dieselgate: Detecting and tricking testers

#42

Post by bryan_w »

Spivak wrote: post #40
cowpig wrote: post #32

Earlier quoted context omitted.

Ah yes let's be sure not to judge anyone for anything they do

You can still judge them evil even if the parent was accurate as to the motivations for their actions. Villains are more interesting when they're sympathetic. You're in the planning meeting discussing this feature, you ask "Hey, are we allowed to do this? I thought stand downs were contractural." and your PM says yes, they got the okay from legal. Now what do you do?

> they got the okay from legal.

Now that I could definitely see happening. I would also want that in writing somewhere.

I guess discovery for the impending lawsuits should be very interesting

Re: Honey's Dieselgate: Detecting and tricking testers

#43

Post by autoexec »

Waterluvian wrote: post #26
bryan_w wrote: post #22

I used to work for an ad tech company (which I know already makes me the devil to some around here), and even I think that they crossed a line with this. A lot of industry terms are coded in corporate speak to make them sound better (think "revealed preferences" or "enabling personalization"), but I would genuinely like to know what the engineers thought when doing design reviews for a "selective stand down" feature.…

> what the engineers thought when doing design reviews for a "selective stand down" feature. Possibly a version of, “I lack the freedom to operate with a moral code at work because I’m probably replaceable, the job market makes me anxious, my family’s well-being and healthcare are tied to having a job, and I don’t believe the government has my back.”

I think you can only get away with that excuse so long as you're actively looking for a new job while also collecting data to turn whistleblower (anonymously if need be) once you have one. Ultimately it falls on the employee to do the right thing or get out because they risk being held accountable for what they do. A replaceable employee (which is pretty much all of them) will be especially vulnerable since they can be thrown under the bus with minimal inconvenience to the company.

Re: Honey's Dieselgate: Detecting and tricking testers

#45

Post by rectang »

Spivak wrote: post #40
cowpig wrote: post #32

Earlier quoted context omitted.

Ah yes let's be sure not to judge anyone for anything they do

You can still judge them evil even if the parent was accurate as to the motivations for their actions. Villains are more interesting when they're sympathetic. You're in the planning meeting discussing this feature, you ask "Hey, are we allowed to do this? I thought stand downs were contractural." and your PM says yes, they got the okay from legal. Now what do you do?

It’s easy, looking at the current state of affairs, to conclude that ethical behavior is incompatible with capitalist ambition. One might still choose to be ethical nonetheless, but with the understanding that you will be overtaken by those who have made a different choice.

Re: Honey's Dieselgate: Detecting and tricking testers

#46

Post by autoexec »

fragmede wrote: post #33
the_snooze wrote: post #4

Original MegaLag video: https://www.youtube.com/watch?v=qCGT_CKGgFE You'd think that if you were an engineer building and maintaing a system like this, you'd have an "are we the baddies?" moment, but guess not.

Capitalism is great at washing its hands of evil. I don't know how much slavery went into making the smart phone that I'm posting this from, but I'm sure it's not zero. I'm ethically complicit in the whole scheme. The C in ACAB stands for Capitalists. Which unfortunately, is all of us.

We're not fully complicit all of the time. You don't know how many slaves made your phone, but somebody does. If you had a choice between a phone you knew was made by slaves and a phone that wasn't I assume you'd pick the slave free version every time. While it's fine to feel guilty for your involvement in the scheme don't let that get in the way of placing the blame for it squarely on the people who set things up this way and put you in this position.

When you can't escape an evil system you just have to do your best within it, while either working to get out of it or working to improve it however you can. What more can anyone ask of you? Capitalism is pretty much inescapable, but thankfully I'm not convinced that capitalism is an evil system inherently, it just needs strong constraints and regulations to keep it from being used to do evil things.

Re: Honey's Dieselgate: Detecting and tricking testers

#47

Post by neilv »

zaphirplane wrote: post #39
Aurornis wrote: post #29

Earlier quoted context omitted.

From my experience, it’s more likely that the engineers who got far enough in the company to be working on this code believed that their willingness to work on nefarious tasks that others might refuse or whistle-blow made them a trusted asset within the company. In industries like this there’s also a mindset of “Who cares, it’s all going to corporations anyway, why not send some of that money to the corporation that…

Do you know of anyone declining to work on a project For ethical in their view ( non military non killing) ? I’ve led a sheltered life and never met one, people have told me they wouldn’t apply for a role with a company for ethical reasons maybe they even believed they would get the job

I know a lot of people who won't work for some companies for ethical reasons.

Though, sometimes the exact reason is muddied, since companies that are perceived as unethical in how they behave externally are often also perceived as unethical in how they behave towards employees. So you might object on pragmatic grounds of how you'd be treated, before you ever get to, say, altruistic grounds.

Also, sometimes fashion is involved. For example, many people wouldn't work for company X, because of popular ethical objections to what they do being in the news, but some of those people would probably work for an unknown company doing the same things, without thinking much about it.

But often it's just "I don't like what company Y is doing to people, and I wouldn't work on that, even if they treated employees really well, and it was really fashionable to work there".

(See, for example, the people who refused to work for Google after the end of Don't Be Evil honeymoon phase, even though they generally treated employees pretty well, and it was still fashionable to work there.)

Re: Honey's Dieselgate: Detecting and tricking testers

#48

Post by bedelman »

ZoneZealot wrote: post #8
the_snooze wrote: post #4

Original MegaLag video: https://www.youtube.com/watch?v=qCGT_CKGgFE You'd think that if you were an engineer building and maintaing a system like this, you'd have an "are we the baddies?" moment, but guess not.

For context, Ben Edelman the author of the blog post was in the video at https://youtu.be/qCGT_CKGgFE?t=1980 Their personal site is also linked in the video description https://www.benedelman.org/honey-detecting-testers/

Ben here. My real (substantive) write-up is https://vptdigital.com/blog/honey-detecting-testers/ . Happy to discuss / answer questions / etc.

Re: Honey's Dieselgate: Detecting and tricking testers

#49

Post by bedelman »

cwal37 wrote: post #2

Archived link: https://web.archive.org/web/20251230214339/https://vptdigita...

Was the VPT site not working for you, so you had to resort to archive.org? Original link https://vptdigital.com/blog/honey-detecting-testers/ . Anyone having trouble -- contact Ben Edelman (easily found by web search) and I will genuinely value the opportunity to get to the bottom of what is wrong.

Re: Honey's Dieselgate: Detecting and tricking testers

#50

Post by bedelman »

choult wrote: post #17
delusional wrote: post #16

Likening any of this to Volkswagen emissions compliance scandal does a huge disservice by treating "Affiliate Marketing" as far too important. "Who gets a kickback on this toothbrush" is a much MUCH less important question than "do you pollute the air we are all breathing".

It's comparing Honey's behavior to a well-known and comprehended scandal. Simile is a tried and tested way (hah!) to explain otherwise potentially hard to understand or dry content. It's not about the severity of the impact, its the fact that they were breaking the rules and explicitly coding to actively avoid being caught by testers.

choult: The factors you mention are the factors that led me to propose the "Honey's Dieselgate" title and to compare Honey to VW.

Of course I agree that health is more important than affiliate commissions. So the comparison only goes so far.

Post reply on HN