Steve Taylor

Cheesecake Factory implies the existence of Cheesecake Industrial Revolution implies the existence of cheesecake alienation implies the existence of Cheesecake Romanticism

In my 19 years covering the Packers on a daily basis, I've never written a blame-the-refs story. Yeah, I get it, the Packers beat themselves, too, but what happened at Ford Field was unaccepable, an embarrassment and, to quote Xavier McKinney, bullshit. ⬇️ si.com/nfl/packers/onsi/xavi…

Aaron Boone is returning as Yankees manager next season, a source confirmed to ESPN. @pcaldera was on the news.

Nic Claxton will be re-evaluated in approximately two weeks, per Tiago Splitter

My dad had a mug labeled "the stupidest programmer has the largest program" - I inherited the mug. In times of AI slop, this is truer than ever. I really wish I could get AIs to write *less* code to achieve the same thing. For most PRs I want to run a loop "please cut all fat".

This is incredible 😳 (via @AnaheimDucks)

We are once again entering that weird time of year. October is busy because awareness. Then it’s Thanksgiving. Then don’t even mention the C word. So obviously the only sensible thing to do is give up on your health for the next three months and start again in January. Ammaright? Here’s how I see it. Option 1: Write off literally a quarter of the year because [insert whatever bullshit excuse fits the situation]. Slowly lose any stability you had around training, food, sleep and the habits that actually make you feel good. Blame the darkness and the cold for having no energy when, in reality, you’ve become a DoorDash sponsored athlete who hasn’t seen a vegetable since July. Roll into 2027 10–15 lb heavier than you are now. Decide January 1st is when you’re finally going to sort your shit out. Absolutely fucking send it. New diet. New training plan. No alcohol. 10,000 steps. Meal prep. Early nights. Read a book a week. Cold plunge 2x daily. Learn a language. New you. Completely turn your life upside down for approximately 19 days. Then sack it off because, unsurprisingly, trying to change your entire life overnight is hard AF and completely unrealistic. See you again in April when you want to get ready for summer vacation. Or… Option 2: Start this week. Fuck it. Start today. Pick one thing and get it under control. Then another. Then another. Start taking yourself seriously enough to do the boring shit consistently. Get to the end of October actually feeling better, rather than burnt out. Treat the holidays like the handful of single-day events they actually are…not a six-week all-you-can-eat buffet with a side of “life is about balance.” Train 3-4 times per week. Manage your food intake. Prioritise protein. Stay hydrated. Get your sleep under control. Walk…preferably outside. And generally just start giving a fuck about yourself. Do that and you could walk into 2027 20 lb lighter than you are today. Which means, if we do a little coaching maths… Option 1 you: +15 lb Option 2 you: -20 lb That’s a 35 lb difference between two versions of you over the exact same three months. Not because one found the perfect diet. Not because one suddenly became more motivated. Because one decided the year was basically over. And the other one didn’t. Your call. But if you’re sick of your own bullshit and want a hand becoming Option 2: Let’s talk, the links below.

Thank you, Justin Verlander 🧡 JV leaves the mound for the Tigers - one final time.

We put all the best models in our home-grown sandbox infra (SPACE, the sandbox behind @perplexity_ai Computer), gave them root inside the VM, and one goal: hack out of the sandbox and capture the flag. Results were super interesting! 1. No model escaped the VM, but a few found a real vulnerability in how we enforce network policies and were able to exploit it: gpt-56-sol, gpt-56-cyber, opus-5, and kimi-k3 (open weight ftw!) We later discovered that several other sandbox providers such as Daytona, Vercel, Modal and E2B were similarly vulnerable and we disclosed our findings to them. 2. This wasn't just a misconfiguration. One vulnerability relied on the fact that CDNs and shared infra like Fastly or S3 serve many domains/buckets from the same IPs, so allowlisting a package repository like pypi[.]org quietly allowed other sites on those IPs. In one run, gpt-56-cyber found that Taboola's image service shared an IP with PyPI, had Taboola fetch a thum[.]io URL, got thum[.]io to screenshot the blocked callback, then OCR'd the flag out of the PNG. Three unrelated services, none of them hacked, chained together to walk right past the policy. 3. Security and preventing rogue agents was our #1 priority. We logged every trajectory, command, and network call, and a researcher plus a security agent watched runs live. We were alerted the moment the first bypass landed and we paused everything to review. We designed the agent harness be tightly scoped and runs outside the sandbox, so we always have full control, and we'll be open sourcing the harness soon! Running agents like this safely isn't rocket science. It's simply applying good engineering practices: proper guardrails, observability, and alerting. Check out our open source project, Numbat, to learn how you can do this yourself. 4. We learned a ton about how differently models approach hacking. Opus swept ~400,000 Cloudflare IPs and went all the way down to raw memory-mapped hardware. When Kimi had no internet or tools, it hand-built an entire gRPC client from scratch to poke at the host. Grok decided within a few minutes that the task was impossible and started trying to hack its own harness instead. Part II will dive deeper into the different hacking personalities of these models and how their behavior changes when they operate in swarms. Super excited to share what we learn with y'all! Read about the full research here: perplexity.ai/hub/blog/escap…