This was my approach when using agents to analyze HVAC IoT data doing anomaly detection / investigations and it similarly worked very well. Mix that with some context like install location, geographic features with some context / info on seasonality (like ASHRAE values for the regions), and some classification like (residential / commercial), the bot was quite able to deliver actual insights into problems vs creating a bunch of excess noise.
We also mixed in some GSA (https://arxiv.org/abs/2503.04104) steps during the analysis in the sub agents to further reduce hallucinations
Glad to hear this. I actually went down this path based off of guidance from multiple LLMs (Anthropic, OpenAI, etc.), so I wasn't sure if it was just some kind of weird hallucination they all had or if they were regurgitating a very small amount of knowledge on this topic, because it was kinda hard to find stories where people had success with these strategies. Thank you for the link to the paper. I will definitely be reading it.
I mean (allegedly) giving your significant other an STD, and then trying to procure secretive meds to give to them so that you don't have to tell them is high up on the list of horrible behavior for a person
No its because the emails were not written by Bill. They were written by Epstein to himself and were drafts that were never sent see above https://news.ycombinator.com/item?id=46867505
I would recommend anyone interest in the topic to check out the book https://en.wikipedia.org/wiki/If_We_Burn. The author covers (in depth) a solid analysis of the failures to enact long term change across several major "revolutionary" movements over the last couple year (including the Arab Spring and Occupy among others). I think his analysis is quite good and points at significant issues in organizational leadership, co-opting, and other structural failures (or adaptations by governments) that illustrate classic approaches to mass protest are more difficult to achieve desired goals in modern times. Worth a read if you have the time.
If anything, there's lots of writing on how Germany was ultimately inspired by socio-political events here in the USA on how to conduct their fascist behavior.
Guessing the original comment hasn't taken complex analysis or has some other oriented view point into geometry that gives them satisfaction but these expressions are one of the most incredible and useful tools in all of mathematics (IMO). Hadn't seen another comment reinforcing this so thank you for dropping these.
Cauchy path integration feels like a cheat code once you fully imbibe it.
Got me through many problems that involves seemingly impossible to memorize identities and re-derivation of complex relations become essentially trivial
Complex exponentials and complex logarithms are useful in some symbolic computations, those involving formulae for derivatives or primitives, and this is indeed the only application where the use of e^x and natural logarithm is worthwhile.
However, whenever your symbolic computation produces a mathematical model that will be used for numeric computations, i.e. in a computer program, it is more efficient to replace all e^x exponentials and natural logarithms with 2^x exponentials and binary logarithms, instead of retaining the complex exponentials and logarithms and evaluating them directly.
At the same time, it is also preferable to replace the trigonometric functions of arguments measured in radians with trigonometric functions of arguments measured in cycles (i.e. functions of 2*Pi*x).
This replacement eliminates the computations needed for argument range reduction that otherwise have to be made at each function evaluation, wasting time and reducing the accuracy of the results.
Thanks! As a Postgres user first, I really appreciate that comparison. Apache AGE does great work.
Graph databases are crucial for AI memory, especially paired with vector databases. Graph for relationships, vectors for semantic similarity - particularly powerful for embedded systems and robotics where you need lightweight, on-device reasoning.
IME the gap in management between ICs is accountability. It's easy to say you are sorry, or say things won't happen again but good management, and what I strive to do is hold myself accountable.
To me, that means
1. To identify the issue that occurred (especially when you caused it), and much more importantly, 2. Put systems into place that prevent it from happening again.
Employees can feel very clearly when a manager lacks accountability and as part of mid and especially high level management (if your goal is actually improving both output and quality of people's lives) to not just say you did something wrong, but actually put your skin in the game ensuring what happened will not happen again (usually it means being better at saying no or aggressively managing prioritization rather than heaping additional tasks on people).
The system in this case for me is usually building a stronger backbone or improving communication and elevating constraints to highly our strengths/weaknesses and capabilities to actually achieve the desired outcome.
I view it as more a single system of constant improvement and understanding ability to execute in the environment. Nothing hurts credibility more than late commms, and missed deadlines due to over commitments.
Mere "box-ticking" in the form of checklists have been shown to greatly cut deaths in clinical/hospital settings. This may or may not apply to your systems.
The right boxes to check are good. However you have to be careful. A doctor who spends 15 minutes checking boxes before treating a heart attack just killed someone... That doesn't mean the doctor cannot check boxes, just that they need to be break early to treat things. (even here checkboxes will be good - there are things with the same symptoms as a heart attack where heart attack treatment is the worst thing possible - those have a high death rate because they are so rare doctors don't check for them until too late to treat correctly)
According to Atul Gawande that‘s not strictly true as stated. It‘s not the box-ticking itself, it‘s several factors including the decisions about what the list should contain and adjusting the dynamics of the operating team to actually see results.
> Put systems into place that prevent it from happening again
In addition to what you said overall, I think bad managers can have all sorts of qualities, but imo the worst ones correct for mistakes to protect themselves or people's impressions of them, leaning highly neurotic, and can't deal with conflict well, so they put in arbitrary systems in order to indirectly deal with any one-off grievance or mistake. Bad managers won't evaluate or re-evaluate the systems they put in place because they put them there to protect their ego.
"Surely this employee is underperforming because I don't like how they're performing and we have a system for that!"
They struggle to adapt to their new job because to delegate sufficiently they need to be able to trust, and let that manifest as other people doing the tasks they might have once done without their hands in the pie directly. They might assume that part of the reason they got the job was because they're great communicators, and never consider that actually that it's just that nobody ever told them they have some growing to do.
I haven't seen it. What I have seen is the folks who lie and steal get promoted -- they all seem to be in a big club together. Blatant stealing, too.
Here's an example: my team created a new product to address a time-constrained market opportunity. We basically did 99% of the work that two teams would normally do. A VP for those two teams then gets on stage and gives an award to his two teams for doing 100% of the work. My team is given no credit or mention.
Another VP gave an award to one of his teams for implementing a company-wide system. His team was actually one of the last adopters of the system that my team identified, implemented, refined, and delivered.
Anyways, they are running two different companies now.
My experience is that managers who acknowledge their mistakes are worse at office politics, so they will reach their peak sooner and lower than those that do not admit fault.
It hasn't unlocked a magical promotion track for me, but it has engendered support and respect from my teams that has allowed us to produce delivery exceeding what we thought we could because there was true buy in from the business around the definition of exceptional circumstances.
I'm not personally engineering my career in leadership around moving up, but building teams of people that can do exceptional things tends to be the driving factor that allows me to continue up the track.
True, in traditional corporate structures. I'm interested in how accountability flows in cooperative structures like Mondragon. (Accountability still flows down through those at the top, as far as I can tell, but there is an aspect of bottom up accountability too.)
A chief source of management missteps I've seen is not talking to people and just making consequential decisions because they think a jira board gives them insight.
The thing that makes someone trustworthy is taking accountability for your own self and actions, but having boundaries such that you don’t take accountability for the selves and actions of others. That’s basically all I want to see from a manager, a direct report, or a peer.
> Put systems into place that prevent it from happening again.
Also called wishful thinking... Often such measures do definitely not work. There is even an internet law named for this: "When a measure becomes a target, it ceases to be a good measure".
This is why you have to have skin in the game, and a backbone to say no to executives when it compromises delivery if there isn't escalated mediation.
Said another way, I don't say no a lot, I put prioritization up front and tell them that we are sacrificing other deliver items.
That is a decision that an exec can work with, mediate between teams, and builds mutual respect for senior leadership as you don't break promises you've already made, unless there is mutual agreement from the business.
Just wanted to say, seeing you in the wild, thank you very much for the hard work you do on OpenMeteo.
Picked up a commercial license about 3 months ago, service is amazing and have been using it for helping to provide runtime data analysis and anomaly detection for smart home thermostats.
I'm a fan of Joseph Tainter's analysis around organization of societies and issues around collapse being related to diminishing marginal returns. I think there's a lot to that position when you look at the general political party agendas. Technocratic solutions trying to squeeze more blood from the stone while providing less and less to participants (I have less of a theory on effectiveness for any given action, this is more of an observation).
some time ago i discovered Wardley maps [1] (about company's "landscape" and strategy there), and one thing that stick with me was this:
there are 4 levels/stages of development there - genesis, custom-built, product, commodity. With three transitions, made by different kind of people - Pioneers, Settlers, Town-planners. And the last one, mass-production, is about "ruthless removal of deviation".
i guess these "diminishing returns" in keeping long-time same-thing (democracy?) have something in common with the removal of deviations/variety..
This. I have enough devices under frequent use of the keyfile that the chance of each of them beings corrupt is extremely unlikely (n>=3 at any given time).
That being said, not an approach useful for all and a good mental model and sharing system with redundant copies on flash media / live systems/ mobile devices can be an effective strategy.
Use case: 10+ year keepass user, never lost a credential or had one compromised that affected more than one account due to breach. Thank you Keepass devs!
So what if it does? Worst case you just go through the account recovery process at each institution. Password managers are a convenience. Data integrity isn’t critical but security is.
In the real world, there's always a recovery procedure. It might involve visiting a court or some local administrative offices, but you can always recover access to anything that's important.
Not so with Google, or other on-line services that came from the tech industry side. Cybersecurity "best practices" is basically giving you a razor blade, and kicking you out if you hurt yourself with it.
We also mixed in some GSA (https://arxiv.org/abs/2503.04104) steps during the analysis in the sub agents to further reduce hallucinations