What are your thoughts on skill-specific cred/how did you plan on implementing that @decentralion?
My simple solution would be to just build out these skill trees then let people assign points to their skill trees themselves. Eg. 10XP = 1SP (skill-point)
Sure, people can lie and pump the skills they werenât working on,
but in the beginning theyâd be disincentivized to do it because others would see,
and later theyâd be disicentivized because taking up a work contract on the market might require staking.
this skill tree is like an online school then and by learning you complete quests? you always get a quest for what you next want to learn so that you can teach yourself a new skill while doing a quest if you want to learn a new skill. The accomplished quest are the proof of your skills.
i like this analogy. like 1 ring for one year experience. the trunk represent the total years of experience of all your skills together and each skill branch has its own thickness based on the sub-branches.
the more flower the more quest s/he did on this skill. so if a branch has no flowers you see s/he did no quest. lots of flowers lots of quests.
The roots of the tree are missing. in nature they get resources from the soil and shrooms. perhaps the roots show the connections to other users/projects. so the more quest you did for a project the more roots/connections you have to this project. projects are represented by other trees (project trees = needle trees & user trees = leaf trees) and Mykorrhiza (shromms) connect all trees. The big Mykorrhiza = MetaGame
this skill tree is like an online school then and by learning you complete quests? you always get a quest for what you next want to learn so that you can teach yourself a new skill while doing a quest if you want to learn a new skill. The accomplished quest are the proof of your skills.
In the long term, hopefully, yeah.
I wanted to start it really minimalistically, by simply mapping out different branches of knowledge one can get competent in without going to university, add snippets of advice & linking the knowledge nodes to external courses, books & other resources.
Then, with time, we may build our own resources, organize workshops & learning sessions, have other organizations tag work that they need done directly to those knowledge nodes & boost them etc.
Weâve thought about it, yeah. Basically you could think of it as âcolored credâ, where there are different kinds of cred flowing through the graph: documentation cred, implementation cred, research cred, etc. This isnât something thatâs moved far beyond the daydreaming stage, though: there are some open research problems weâd need to solve, and it would be a really big change overall.
My 2c: If you want the skill tree to be meaningful you need some regulatory mechanism that audits or verifies peoplesâ skills. Going on self reported skills or on weak metrics can easily lead to the skill levels becoming meaningless:
ppl who are actually very skilled (and obviously so) will not bother putting much effort into pumping their stated skill level, they already have validation from themselves and their peers
ppl who are mediocre (or dunning-kreugered) will try a lot harder to get high skill levels, will want the validation more
Once youâre in that situation, ppl will get burned a few times by players with high skill levels but a shortage of actual talent, and then ppl will learn to ignore the stated skill levels.
I think you need some kind of social regulatory process (maybe adversarial). Some examples of real world processes that achieve this:
You get a PhD after successfully defending your thesis (note |he adversarial framing) from other people who have already been proven by this system
in Google, you get promoâd by a committee of people that are at the level youâre trying to get promoted to (or higher)
If we wanted to do it with existing XP/SEED mechanics, we could maybe do something with boosting where to get promoted to a higher skill level, you need to collect enough seed power in boosting that supports your candidacy, and then it gets approved or rejected by a review group of people who are already at that skill level. The boosters would earn XP if they boosted a successful candidate (but lose their SEEDs with no recourse if they boosted a faker). So they would actually be incentivized to talent scout people with high skills. Just a very back of the envelope idea, I think we could expand it into something interesting. cc @METADREAMER
I thought about this a lot, because I thought it was important that the people with high skill that want to join donât have to go through all the hoops of verifying skills which wouldnât be exactly cheap for us either.
My conclusion was:
You are free to claim your skill level to be as high as you want (which unlocks you access to higher skilled work)
BUT! If you try to take a high skill contract, youâre going to have to stake to collateralize your rep or lack of thereof.
If thereâs disputes about you being incompetent, you get slashed (and a bad review on your skill tree)
Will self-asserted skill correlate with actual skill?
Dunning Kreuger effect says: systematically, no. People with low skill will overestimate (they donât know any real experts, donât know how much they donât know), people with moderate-to-high skill will underestimate (they know enough to see that they know far less than the masters)
People who have less to lose reputationally (e.g. new to the community or short-lived accounts) will be willing to gamble on inflated skill claims
Will slashing reliably correlate with being low-skill?
Many people are naturally conflict avoidant and will not slash in questionable situations
People who are good at playing politics / reading social situations less likely to get slashed
A system where people need to post collateral to claim contracts is discouraging â what if I have a lot of skill, but not a lot of money to risk on the possibility that the person posting the contract is an asshole? Or the contract poster may have unrealistic expectations and zero experience with product management, and will blame all those failures on me.
Overall, I think this system will see a lot of highly inflated skill scores, but fear on both sides:
Contract poster worried that the skill claim is inflated. Doesnât want to risk investing energy and resources on giving the contract to someone based on that flimsy metricâeven if itâs possible to slash, that may not make up for the lost time and energy
Contract taker worried that they may get blamed/slashed when they are actually doing a good job, but the person posting the contract had unrealistic expectations or other problems
Iâm familiar with Dunning Kreuger effect, but things like that can be overcome simply teaching people how to self-assess and later improve it with other signals.
Sure, but if youâre trusted enough, youâll be able to have other people or agencies stake for you.
And it goes without saying, the person on the buying side =/= the arbitrator.
I think it depends on how much social cohesion/community policing you expect on this system.
If itâs a small-world network (most ppl know each other, community around the size of Dunbarâs number) then I think background social feedback could keep the system in check.
Once it scales to >1000 people though, I think there needs to be a more robust framework for assessing skill levels.
I think having âguildsâ that assign skill badges could be interesting. Different guilds can experiment with different mechanisms for ensuring the scores are meaningful, and guilds can compete with each other to do a good job of providing useful signal.
Yea, RaidGuild has been talking a lot about an apprentice > journeyman > master progression system.
Iâm not a fan of self selection either, donât think it will scale or have a lot of weight in practice. Might be worthwhile to make the first version super simple, something like LinkedIn does where people can select skills on another persons profile for things they are good at. Just simple votes, but can also weight them based on the reputation of the people giving those votes.
The only thing I think could be self selected in a reasonably objective way would be âyears of experienceâ in each domain. That sort of anchors things to an objective metric: time. Of course years of experience is not a perfect representation of skill level, but its still a pretty good metric and probably the only one that could hold some sort of weight. Of course people can lie about it, but its still less prone to dunning-kruger IMO. Use that as a basis, and then have the voting mechanism on top.
Honestly the best way is just for people to fill out their bio properly to just talk about stuff they did, and then people can judge for themselves based on their work and portfolio how good they are.
Thats nice. I like it a lot. It looks like a city/brainmap very organic, but at the same time very chaotic. Then you can activate filters like âshow my skillsâ or you can click on a branch and see all the possible path from this skill or you can click âshow next possilbe skillsâ and you see all the skills you can learn next.
It is an insane tree (but only in terms of layout) because the underlying attack/def can be derived from simpler rules and segmentation of classes. Iâm reminded of the old quote
some things that count canât be measured, and not everything measurable counts
Back to your skill tree ⌠your vertical ovals represent knowledge domains ⌠and horizontal the current state of expressive language. Accounting has their own language and so do lawyers with their shorthand jargon ⌠itâs all to speed up communicating high-level constructs / patterns that are observed externally. Communications is underrated as you need to either translate, bridge or laconise the key points otherwise you risk putting audience to sleep or trying to cram TLDR by starting from fundamentals.
The other interest observation is that people that rise to the top tend to have A-shape or T-shaped skills with legs in two or more knowledge domains (eg me in HPCC and IP law). Others might be T-shaped in rising through technical ranks, then after a certain point forced to go sideways to get to C-level. What we should be looking for in cryptoeconomics is people with H-shaped, a leg in one of engineering+biz/econ, and now want to cross-over before rebuilding again.
From my perspective, what is hard to do is get cross-functional teams that mesh well, either because of
values conflicts (never ask accountants for opinions on research budgets),
lack of common framework, or
simply just culture clashes (eg intergenerational).
The metagame can attempt to bridge some of this by amplifying some traits ⌠eg the skills in internal support ops need some refactoring before being turned into external outreach conceptualised in going from necromancer (using RaidGuild term) to water mage (customer data analytics).
I like the idea of mycelium and soil included in the skill tree. Maybe skills beneath the trunk would be about connecting with others. Like mycelium helping trees to communicate what nutrients to pass between themâŚ
@mzargham wrote more about it in MetaSkillz & pushing forward the technical side with the pre-requisite - MetaPubz
The Deep Work team is also working on implementing a version & Raid Guild supposedly too but I donât know whatâs the status. Probably a bunch more too.
On our side, weâll be pushing it forward at MetaFest - there will be a workshop about building it followed by a session of mapping skills in Miro.
Also, I watched the deepwork video in the post above and I recommend people reading this thread check it out. I added it to the list of linked resources in the hackmd above as well.
âOh shit. Sorry I just burnt my food.â Lol.
Yes that was a nice clean video. The trajectoTree looks digestible. The side tables are nice to see how working groups are structured. Thank you