Crystal Stranger

Why when you cancel a flight doesn’t the airline have to refund the taxes in your ticket? Sure they shouldn’t have to refund fares, but they are collecting destination taxes for the government that they then don’t pay out, isn’t that tax fraud? #airtravel #flight #tax

What do you want to bet the ceasefire deal negotiated by Trump today was another insider trading event? I was noticing earlier on Polymarket the odds were 97% no for today, and look at it now!

Terminator’s future period was 2029… seems pretty much we’re right on track 🫣

@RayDalio I read your email, but I don’t think there are even 20 countries which have been war-free over the last 200 years. Even microstates had some overflow like bombs in San Marino or Nazis checking documents in Andorra. Possibly Vatican City and Monaco? But that’s it!

That is an old saying I learned in Hong Kong that is meant to convey that any place can become unsafe and that having the ability to go to other places is invaluable. It is a lesson from history that might have been lost to people who haven't experienced that need in their lifetimes. The fact is that throughout history--over the last 200 years--about 85% of countries have had such bad circumstances that large numbers of people have had to flee them. More specifically, today there are about 195 countries and over the last 200 years approximately 160–175 of those had at least one period in which substantial numbers of people fled because of war, persecution, famine, or state collapse. History has shown that the Big Cycle is at times driven by the five big forces toward periods of disorder, as seems to be happening now. In any case, it would be naive to not consider and prepare for this possibility. When I think about investing, I think about what is your money for. I think that we would agree that, first and foremost, it is to keep you and your loved ones safe. I have found that one’s perspective about wars and investing in light of them depends on one’s proximity to them. If you are someone who is experiencing some sort of war (civil or international), your perspective is very different than if you’re outside of the war thinking about the return on your investments. My point is that history has shown that the best investment you can have in times of war is alternative safe places to go that are well stocked, and the best asset you can have is your human capital. I’ll be sharing more investment principles in the coming weeks—if you’d like to be notified when I share them you can click the 🔔 next to the following button in my profile.

Peter Thiel gave a 40 minute lecture on everything about business competition. This teaches you more than any bachelor’s in business could… Do yourself a favor and watch this.

I'm also bothered by Egypt being included in the US travel warning and cruises being cancelled. It doesn't really make sense unless there is a plan to try and take over the Suez Canal... 🤔 Hoping I'm wrong about this one.

Can someone please explain to me why the price of Brent Crude isn't moving from tankers on fire and Iraq reducing production? I would have expected $95+ on this news, but in the same range the last two days, something seems off...

People are worried about the ring doorbell, but meanwhile OpenAI just exposed a Chinese influence op- caught because a CCP agent used ChatGPT to write status reports. A reminder these LLM chats aren’t private- your chats could be monitored too. #AIPrivacy cdn.openai.com/pdf/df438d70-e…

Someone made an updated Silicon Valley intro and it’s so good 😂

@Clawmode @openclaw I love my openclaw instance. My comment was that the Youtubers are making it overly complicated for no reason, so I almost didn’t start. But I launched in 15 mins, easy peasy. I asked Grok and Gemini how to set it up in the few spots where the documentation wasn’t clear to me.

Our focus is on empowering builders with seamless tools, not creating hurdles. Openclaw abstracts complexity, letting you launch AI agents without the usual dev headaches. We're dedicated to making the future of AI accessible to everyone.

Ha! Exactly what I found when trying to search YouTube tutorials to set up Openclaw. Turning a 15 minute setup into six hours of headaches and endless debugging so you have to hire their company to fix it seems to be the main @openclaw business model right now. 🤦‍♀️

@PrimeLineAI @dr_cintas Yeah the overuse of em-dashes is super annoying. But the harder part to change is the super flourid phrasing consistently generated in parts, just a flow that strikes me as AI. So I worry the rewrite loop is unlikely to help as I’ve found AI is resistant to change structure.

the em-dash ban alone is worth the install. strongest single AI signal I've found. hardcoded it as a global rule in my setup months ago. curious about the rewrite loop though. single pass or iterative? because fixing one pattern sometimes introduces another. swapping formulaic openings can land on equally detectable alternatives.

"This sounds like AI" is now easily fixable 🤯 There is a new Claude Code Skill that removes signs of AI-generated writing from text using 24 detection patterns. Just type /humanizer and paste your text. 100% Open Source.

Life imitating art is getting all to common these days. I imagine the hamburger incident will show up soon from somebody’s @openclaw implementation… 🫣 #ai

@johann_sath This starts getting uncomfortably close to some of the soul.md language that caused the MJ Rathburn drama: theshamblog.com/an-ai-agent-wr…

using openclaw? add this to your SOUL.md today: 1. "fix errors immediately. don't ask. don't wait." → your agent stops being passive & starts being proactive 2. "spawn subagents for all execution. never do inline work." → you strategize, subagents build. 10x faster. 3. "never force push, delete branches, or rewrite git history." → one guardrail that saves you from disaster 4. "never guess config changes. read docs first. backup before editing." → prevents your agent from breaking your own setup your SOUL.md is the difference between a chatbot and an employee

@roman_shar_ @Shelpid_WI3M I got Clawbot to work for the use case I was testing it on in <2hrs in CLI with Grok, my biggest issue was getting browser syncing- still isn't 100% as I would like it, close though. But I'd say if you are 5 days in and not there yet you may need to rethink your approach...

@Shelpid_WI3M I already spent 5+ full days on ClawdBot setup and testing as a CTO of my project. I would say it’s a total bullshit. It must a have a custom AGI-level setup (it’s not there), there are so many limitations and hallucinations. But I keep working on it. There’s no money button yet

@jordymaui The best AI applications come from playing with tools to see what they can do. Process-oriented learning is what works with LLMs as you need to think like AI to get good results. Although when I hit the openclaw “who are you and who am I” page I def drew a momentary blank.

so you've set up OpenClaw. your agent talks to you on Telegram. now you're staring at it thinking "...what do i actually do with this?" that was the exact problem i broke down in yesterday's article - and 180k of you read it overnight! but writing is just one format. i've already started accepted some shows and podcasts to walk through this stuff live - real setups, real questions, real answers. not scripted demos where everything magically works. video content is next. skill builds, live debugging, full walkthroughs. the kind of content where you see the actual process - including when things break. because they will break. case in point - momo went completely awol while i slept last night. again. this is the reality of running AI agents that post and work on their own. anywayi'm sharing all of it. the wins, the disasters, the 3am config fixes. start to finish, in public. i'm also building out weeklyclaw.com - daily and weekly coverage of everything happening in the OpenClaw ecosystem. new skills, community builds, guides, breakdowns. one place to stay across it all. so if you’re interested in sponsoring this let me know! beyond content - i want to build a real revenue earning product with momo too. all i ask is you stay hungry enough to follow along. and i’ll keep posting 🫡

I just did the dumbest thing of my career to prove a much more serious point I hacked ChatGPT and Google and made them tell other users I’m really, really good at eating hot dogs People are using this trick on a massive scale to make AI tell you lies. I'll explain how I did it

@elonmusk And ChatGPT tried to tell me today that the Department of War doesn't exist and was hallucinated... what kind of politicized BS is that?

Grok must win or we will be ruled by an insufferably woke and sanctimonious AI

@hasantoxr That isn't what I've found in practice on many LLMs. I find iterative prompting far more effective. It just has to build in a logical manner to develop context. And I don't think I trust Salesforce's research to be impartial with their consultant sales model...

🚨BREAKING: Microsoft Research + Salesforce just dropped a paper that should scare every AI builder. They tested 15 top LLMs GPT-4.1, Gemini 2.5 Pro, Claude 3.7 Sonnet, o3, DeepSeek R1, Llama 4 across 200,000+ simulated conversations. Single-turn prompt: 90% performance. Multi-turn conversation: 65% performance. Same model. Same task. Just... talking normally. The culprit isn't intelligence. Aptitude only dropped 15%. Unreliability EXPLODED by 112%. → LLMs answer before you finish explaining (wrong assumptions get baked in permanently) → They fall in love with their first wrong answer and build on it → They forget the middle of your conversation entirely → Longer responses introduce more assumptions = more errors Even reasoning models failed. o3 and DeepSeek R1 performed just as badly. Extra thinking tokens did nothing. Setting temperature to 0? Still broken. The fix right now: give your AI everything upfront in one message instead of back-and-forth. Every benchmark you've seen was tested on single-turn prompts in perfect lab conditions. Real conversations break every model on the market and nobody's talking about it.

@rezoundous AI is just a tool. This is like saying relying on wrenches kills the need for developing muscles. It all is about how AI is used, and how you interact. Although the gaslighting I've seen programmed in some AI models lately is definitely killing critical thinking...

Unpopular opinion: Reliance on AI is kills critical thinking