author: niplav, created: 2023-07-17, modified: 2024-09-12, language: english, status: finished, importance: 5, confidence: other
My services (research, programming, experimental design and implementation) can be acquired for money.
I am available for hire to do consulting, especially research consulting and to a lesser degree software engineering and data science.
Contact me here for inquiries.
Examples of my past paid work include a library of forecasting datasets an analysis of range and forecasting accuracy, and a literature review on transfer learning in humans. See also my track records.
My rates are determined through a two-sided sealed-bid auction, splitting the difference in bids1:
$r_{\text{min}}$ per hour.$r_{\text{max}}$ per hour.$r_{\text{min}}$ and $r_{\text{max}}$, either first through hashsums of the numbers (with random text appended) and then the cleartext, or simply at the same time.
I am also open to implementing the VCG
mechanism
in a negotiation, which would be more effort (needing to report full
utility functions for salaries, and deciding on $h_i$) but also
more fun.
The current state of the art for salary negotiations is really bad. It rewards disagreeableness, stubornness and social skills, and is just so unelegant.
Me posting fixed rates online is slightly better, but allows me less flexibility: There are projects I think are hugely important and/or interesting, which I'd like to prioritise above unimportant and uninteresting projects, as well as situations when I'm busy, and situations in which I'm not busy. But fixed rates only allow this to a limited amount, and having a complicated multi-stage payment structure is a bit ugly. It also doesn't allow for price discrimination—in some cases I might suspect that most of the gains from trade are going to the other party, and I'm selfish, so I'd like to have some of that :-).
So the system outlined achieves multiple goals: It allows me to bring a potentially better salary negotiation mechanism into the world, in a low-stakes context (since I don't depend on this for my livelihood), it allows for a better distribution of gains from trade, and it allows me to adjust my rates depending on my situation.
Also, $r_{\text{min}}$ and $r_{\text{max}}$ do not need to be
positive! It might be that I like your project so much that I set
$r_{\text{min}}$ to zero or even negative—I have in the past just
started projects other people proposed because I thought the idea was
awesome, and an exceptionally great idea might be
worth paying for. Or $r_{\text{max}}$ might be negative, in that case
you would be selling me something. Because this is a new and system,
I welcome inquiries (silly and serious) here.
I'm not aware of anyone proposing this kind of auction for salary negotiation in particular, Claude 3.5 Sonnet states that it's similar to Vickrey auctions, but in this case there is no second price, and both parties are symmetrical.
I think that the setup described is probably not incentive-compatible due to the Myerson-Satterthwaite theorem, like the first-price sealed-bid auction. (I still think it's a vast improvement over the current state of the art, however). For an incentive-compatible truthful mechanism the Vickrey-Clark-Groves mechanism can be used. If you want to use the VCG mechanism instead of I'm absolutely down to do that, and would try to figure it out for free. ↩