pull down to refresh
146 sats \ 0 replies \ @deadmanoz 8h \ on: Should you use "you" in your LLM prompts? AI
Just a quick search of the various leaked or back-engineered system prompts (https://github.com/asgeirtj/system_prompts_leaks) reveals many occurrences of "you" in the system prompts directly..
If you're self-employed, you don't actually need to contribute. You can choose to make personal contributions voluntarily, but there's no legal requirement to pay the super guarantee (SG) for yourself. It is, however, tax advantaged to make contributions.
Employers must pay the SG for eligible employees, where eligibility is something like:
- Aged 18 or over, generally eligible to receive SG super contributions
- Aged under 18, they typically need to work more than 30 hours per week to be eligible
There has recently been noise and some changes around additional taxes on large balances. I'm sure this noise will only grow louder over time. Eventually IMO there will be some substantial change in terms of additional taxes on large balances - it's simply too large a pool of money for the Aus government to ignore..
I went through the previous iteration of this last year and I can't recommend it highly enough. @ajonas and the team are doing an awesome service "for Bitcoin" with this program (and the previous incarnations BOSS program, Chaincode residency)
I wrote about my experiences as I went through it: https://github.com/deadmanoz/boss-2025-playground
If you're considering doing this, and want to get the most out of it, I recommend somehow freeing up as much time as you can for ~3 months. I was fortunate in that I had left my previous employment a month or so prior for some time off, so had capacity to get stuck in.
I don't own any pearls!
How about Bitcoin Stamps use
OP_RETURN instead of P2MS for their JSON payloads?Hey @Murch, you're right — I thought I had tidied up all the misuse of provably unspendable but I must have missed some. My apologies!
Stamps are not provably unspendable, but they are deliberately unspendable by design. In each 1-of-3:
- 1 key is the "Key Burn" e.g.,
022222222222222222222222222222222222222222222222222222222222222222. It's a valid EC point, but a NUMS point: no one has the corresponding private key. - 2 keys are data-carrying. These may or may not be valid EC points. In Part 1, I show an example where one data-carrying point is valid and the other is not.
The point is not about validating keys - there's been extensive discussion on this, and we know key validation is insufficient for determining whether a point is data-carrying. But we do know that Bitcoin Stamps are deliberately unspendable by design, and we can incorporate this into our analysis.
The other relevant observations are:
- Empirically, P2MS no longer serves its original purpose of multisig custody.
- Bitcoin Stamps is almost exclusively the only user of P2MS since early 2023.
- Since early 2024, Stamps' use of P2MS has been for JSON payloads only.
I am not advocating for touching existing P2MS outputs, confiscation, or anything of the sort.
I'm simply presenting how data-carrying protocols have leveraged P2MS, their design decisions, and what the data shows about how P2MS has been, and continues to be, used.
From this, I'm asking:
- Is P2MS serving its intended purpose?
- Is the observed use worth maintaining the status quo, or do we consider P2MS for deprecation?
- And more pointedly, should we encourage Bitcoin Stamps to consider using
OP_RETURNinstead, given that the project's original rationale for permanence and art is no longer being met?
Thanks @Scoresby, I was gonna post this on here today!
It really depends on what you're trying to build.
Frontends and web-applications and basic business logic type stuff, you can likely expect decent results.
More complicated applications, with many moving parts, or specialised domain, or performance requirements, you're likely to hit the edges of the current state of the tooling. And depending on how far you continue in building such a project without much critical review, you WILL end up in a hot mess.
From what you say, "a little Python thing to do this simple task with all my excel sheets", you can probably expect good results.
The worst thing you can do is not try IMO!!
GENESIS