[gen220]

It’s fascinating how random selection plays an essential role in the election. The final phase’s selection isn’t random, but many intermediary phases are.

I think randomness is an underrated tool for introducing a sense of “fairness” to problems with strange power dynamics: where everyone is interested in a fair outcome, but the best outcome for any individual causes some harm to the group.

Salary negotiation is one example. Imagine haggling over the range start or range size, and then picking your salary from a uniform distribution contained by that range. I wonder how the distribution of (un-)satisfactory outcomes of this system that would compare to the one we have today.

It’s also interesting to note certain parallels to structures existing today. The group stage draw for the UEFA Champions League comes to mind.

[taneq]

> Imagine haggling over the range start or range size, and then picking your salary from a uniform distribution contained by that range.

I cannot imagine anyone, ever, being happy with being told "hey you're great but you only rolled a 4 on your salary negotiation so you're $15k out of pocket.

[meekrohprocess]

Yeah, it's tough when that happens. But that is kind of how things go: part of the d20 system's charm is that it successfully approximates the absurd range and randomness inherent in ordinary life.

Sometimes you get a company that gets offended at the prospect of negotiation - that's a 1, although if you get an offer rescinded like that it's probably good for you in the long run.

Sometimes you throw out a ridiculous number for whatever reason and the other side says, "mmmm...yeah, okay." That's a 20. But it's not exactly uncommon to gain or lose 5+ figures because of a random person's random whims at a random time.

It can be hard to reconcile that with the notion of a meritocratic society, but cognitive dissonance is the spice of life.

[notahacker]

There's a difference between the appearance of randomness across multiple companies from employers and employees having different priorities (and levels of need, and mistakes, and luck) and actual randomness as a selection heuristic. Actual randomness as a policy gives the company the worst of both worlds: the roll of the dice is likely to often give a good recruit an offer they won't possibly accept but also ensures they can follow it up with a significantly higher offer to a worse employee. Lots of companies sacrifice more productive employees to keep salaries to a budget or consciously pay above market to improve their staff pool, but not many firms aim to both keep many promising recruits' salaries below an arbitrary figure and are willing to offer well above that figure to some obviously weaker candidates within the same time frame.

[Hammershaft]

And if you roll a high salary your employer has an incentive to get rid of you and replace you with someone who rolled a 2.

[gen220]

I’m not sure... the difference between a 20 and a 2 shouldn’t be too large.

An example of a large range in my head is like 120-140k, a more typical one would be like 120-130k. If you’re good, you could negotiate a bigger range.

That’s an appreciable difference to the employee, but basically a rounding error to the employer.

If you have a high salary today your employer already has an incentive to get rid of you and replace you with a new grad. It already happens.

[Hammershaft]

In aggregate what appears to be rounding errors to the employer ultimately makes up an appreciable difference in expenditures.

> If you have a high salary today your employer already has an incentive to get rid of you and replace you with a new grad. It already happens.

This is a little different as the employer is making a trade off between employee experience and employee cost. Employers don't have an incentive to replace an employee with an equally skilled employee at the same market rate, but when the salary is randomly chosen out of a range, employers do have an incentive to replace high rollers with new employees who are statistically likely to roll something lower.

[gen220]

What you should realize is that you haven’t lost 15k.

Yay the outcome of the negotiation, the lower part of your range is something you’d be happy with, and the upper is something you’d be quite happy with.

The employer meanwhile is offering you a range from what they’d like, to what they can tolerate.

Anything you get in the middle is satisfactory to both parties, with the added benefit that you didn’t have to beg for it and they didn’t have to cave for it.

Randomness allows for some absolution of responsibility.

If you’re shameless in salary negotiations (not a bad thing for you), this system puts you at a relative disadvantage. But the median person finds salary negotiations to be nerve wracking. I think this helps people closer to the median, by allowing them to negotiate a better position without feeling as guilty about it.

[ALittleLight]

"Neat! When's the soonest I can re-apply?"

[C1sc0cat]

"hero point" :-)

[cbsks]

> Imagine haggling over the range start or range size, and then picking your salary from a uniform distribution contained by that range.

I think that it’s important that the last step is not random. I can’t imagine the job applicant would be happy with their salary being chosen randomly.

[shajznnckfke]

From my perspective, that’s pretty much how it works anyway. At the end of the day you’re sampling from a distribution. You can improve the outcome by networking, interview prep, and negotiation strategy but all of those are just changing the shape of the distribution. I would reject the idea of intentionally adding more randomness because the signal is already so obscured.

[jhanschoo]

Mathematicians already know that random selection is a powerful tool in fairness in electoral processes, the problem is with asking the modern democratic citizen and politician to accept a random process.

[prionassembly]

Also ensuring the legitimacy of the random process (which is not really about technical features, but about social consensus): just how much easier is to claim a PRNG (or a rabbit that chooses a box, or paper slips inside cookies) is rigged.

[gen220]

Yep! For a modern day example, there's a famous conspiracy [1] the Champions League draw, which involves pulling balls from a bin, that certain balls are warmed or cooled so that the ball-chooser knows to draw them in a certain round.

They could rid themselves of this conspiracy by randomly selecting the ball-drawer (instead of using a paid random ex-footballer), but they don't. :)

[1]: https://www.eurosport.com/football/champions-league/2015-201...

[a-nikolaev]

What if randomness here also might be a mechanism for possible divine intervention?

[TeMPOraL]

I think it's more likely that "divine intervention" often was a cover for randomness, with extra benefit of granting some legitimacy to the process in religious societies.