Tootfinder

Reading https://www.ideou.com/en-gb/blogs/inspiration/how-to-not-know-simone-stolzoff and wondering if people's low tolerance for uncertainty is why people would rather a confident but possibly wrong answer from a chatbot than trust their own judgement …

How to Stop Fighting Uncertainty & Start Working With It
Learn how to build uncertainty tolerance, avoid the three certainty traps, and lead more effectively when you don't have all the answers, with insights from author Simone Stolzoff.

I've been talking before why money won't solve the burnout problem. But let's for a minute assume that you really wanted to help people maintaining #FreeSoftware by paying them. The problem is that:
1. You have to pay them a living wage.
While all monetary help is appreciated by developers, they need a living wage. Not "that should prevent you from starving to death" but the kind of money that can support a honest (but not lavish) lifestyle: pay the bills, feed your family, cover other living costs such as repairs, clothes, appliances, and let you save enough for future emergencies.
It's simple as that. If you can't do that, they're going to need a dayjob. If they're lucky, it won't collide with their #FLOSS work. If they're not, it will kill them. Or they'll fall somewhere in the middle, slowly burning out until they can neither maintain their projects, nor work.
2. You need to guarantee that the payouts will continue.
People need security. They're not going to stay unemployed, let alone quit their job or turn down a job offer, unless they either have good guaranties or substantial savings (or they're in a really bad shape and wouldn't be able to handle the job anyway). The job market is hell, and people just know that when the payments stop, they may not be able to find a job soon, let alone a good job. Even "passively" looking for a job can burn you out.
So yeah, one-off payments and pinky swears won't do. And it isn't even a matter of whether we can trust you; it's a matter if you'll actually be able to continue paying us. And honestly, I don't really know how to solve that. Perhaps by paying up front, but for how long? Finding a job may take more than a year, finding a good job may be once-in-a-lifetime opportunity.
3. It can't end up being a job.
Perhaps most difficult of all, these payments can't really come with explicit obligations. I mean, that's the whole point: you want to support FLOSS, not turn it into a corporate project. You want the maintainer to remain free and enjoy the work. That is unlikely to happen if their livelihood is now dependent on your satisfaction. And even if it isn't, I for example would still feel indebted to whoever's paying me to do FLOSS, even if they really didn't expect anything in return, and would fall into a spiral of guilt-inflicted burnout if I failed to maintain the software satisfactorily.
#OpenSource

@… great talk. Lots of food for thought. https://cfp.gulas.ch/gpn24/talk/TVNEUB/

Note, this is Cisco, not an AI firm, saying there is a real wave of threats coming from AI exploits.
Guidance for defending in the age of AI-enabled attacks https://www.cisco.com/c/m/en_us/about/doing_business/trust-center/cisco-defending-against-ai-attacks-guidance.html

Agent libOS: A Library-OS-Inspired Runtime for Long-Running, Capability-Controlled LLM Agents
Yingqi Zhang
https://arxiv.org/abs/2606.03895 https://arxiv.org/pdf/2606.03895 https://arxiv.org/html/2606.03895
arXiv:2606.03895v1 Announce Type: new
Abstract: Large language model (LLM) agents are evolving from request-response assistants into long-running software actors: they maintain state across model calls, fork subtasks, wait for external events, request human authority, generate tools, and perform side effects that must be resumed and audited. This paper presents Agent libOS, a library-OS-inspired runtime substrate for LLM agents. Agent libOS runs above a conventional host operating system; it does not implement hardware drivers, kernel-mode isolation, or a POSIX-compatible operating system. Instead, it treats an agent as an AgentProcess: a schedulable execution subject with process identity, parent-child lineage, lifecycle state, a tool table derived from an AgentImage, typed Object Memory, explicit capabilities, human queues, checkpoints, events, and audit records. Its central design rule is tools are libc-like wrappers; runtime primitives are the authority boundary. Filesystem access, object access, sleeps, human approval, JIT tool registration, and external side effects are checked at primitive boundaries under explicit capabilities and policy.
We describe the design, threat model, Python prototype, and safety-oriented evaluation. The current prototype implements async scheduling, namespace-local Object Memory, runtime-integrated human approval, one-shot permission grants, per-process working directories, shell and image-registration primitives, Deno/TypeScript JIT tools over a libOS syscall broker, filesystem/object bridge tools, an injectable Resource Provider Substrate, deterministic demos, real-model smoke scripts, and 123 regression tests at the time of writing. Rather than improving planner accuracy, Agent libOS demonstrates a runtime substrate in which long-running LLM agents can be scheduled, authorized, resumed, and audited without treating tool dispatch as the trust boundary.
toXiv_bot_toot

I keep being reminded that highlighting and saving highlights from a non-fiction book does not really work.
When I highlight, I do it with the current context in mind.
When I extract the highlights, their context is gone and those quotes rarely make sense independently.
Who has a system they trust, works for them,
and are willing to share it?