Aiden 🏴 (@_dr_bonez) header
Aiden 🏴 (@_dr_bonez) profile picture

Co-Founder & CTO @Start9Labs | Hacker | Punk | Agorist | Jain | PGP: B416 C0E6 7B9D F3F1 7184 80AE 7CFF DA41 CA66 056A | Registry: https://teapaw.org/t.co/VWKVOosYMP

Denver, CO
Joined August 2016
Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)Photo by Aiden 🏴 (@_dr_bonez)
Aiden 🏴 retweeted
StartOS 0.4.1 is published, and contains important security updates. Please update as soon as possible.
10
41
4
167
9,511
Aiden 🏴 retweeted
I'm tired boss
176
911
437
8,810
2,305,608
Aiden 🏴 retweeted
Releasing the model weights and technical report of Kimi K3. Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window. New model architecture: 2.5x the intelligence per unit of compute, not just more params. Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale. Model weights: huggingface.co/moonshotai/Ki… Tech report: github.com/MoonshotAI/Kimi-K… Tech blog: kimi.com/blog/kimi-k3
1,537
7,159
2,352
45,860
14,758,729
How would banning open-weights models even work? Option 1: Ban downloading/distributing weights > Seems impossible. We all know how to download things we're not supposed to. Option 2: Ban US-based companies (Fireworks/Baseten/AWS/etc) from serving the weights > So US-based companies have to use Chinese endpoints, therefore sending our data to China + running inference on Chinese hardware instead? This seems worse. Option 3: Ban US-based companies from using the models from any endpoint > US-based app-layer companies will be at a huge disadvantage having to pay 100X token premiums. The cheapest/best AI-native applications will then come from outside of the US. So then wouldn't we end up with Chinese Cursor/Notion/Harvey/etc? > Seems like a great way to crush US app-layer Option 4: Ban use of any application (US-based or elsewhere) running on Chinese models > Also no idea how to enforce this > We end up with Great Firewall, America edition?
14
5
1
60
5,150
Aiden 🏴 retweeted
wow am i understanding this correctly? 1. put chinese models on the entity list. 2. force american companies to buy more expensive ai from american labs. 3. watch the rest of the world use cheaper models with equal or better intelligence. 4. discover that distillation did not magically stop because washington published a list. 5. make american businesses less competitive globally while raising prices for ordinary americans. 6. meanwhile none of this stops chinese ai labs from becoming better & better. congratulations you protected domestic ai companies by taxing the competitiveness of the entire country. what a colossal mess. someone please add order to this madness.
We support open-source AI and the innovation it unlocks. But open source is not open season on American IP. When PRC firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations will be on the table.
280
578
62
6,142
697,098
Replying to @drydenwtbrown
Are they copying IP? Almost certainly. US labs have done similar, and did it first. One of many demonstrations of the futility and adverse consequences of making language “property.” Regardless, the IP issue is muddy and not as holy vs evil as you cast it. Your projection that it’s somehow “vad for America” is the more dubious aspect here though. 350 million Americans benefit from cheaper intelligence A handful of trillion dollar tech companies don’t Why do you portray as if the latter is America? Again, more nuanced than your red team blue team analysis. Also you’re seriously underestimating the intellectual work being done in Chinese model companies.
4
4
2
164
7,728
Aiden 🏴 retweeted
One problem with "you shouldn't review the code" is that when I review the code, even with the extremely thorough agentic coding system I have set up, I still regularly find things that should never ever ever ever be shipped, even though it technically works. I don't know when I will be able to ship sight-unseen, but that time is nowhere near right now.
103
56
24
1,252
263,912
Aiden 🏴 retweeted
35
18
3
174
8,501
I'm here for @PortlandHODL's redemption arc. I've heard enough of his spaces to give him the benefit of the doubt, unlike most of the people egging him on in that infamous clip. I think PH's apology was genuine, so I forgave him a long time ago. Now that he's proposing a spam solution, I think it would be a disservice to ad hominem attack the BIP he proposed. I'm not technically proficient in Bitcoin internal stuff, so I'll let the mailing list determine his proposal's merits. From a glance, it is closely aligned with my personal belief that individual transaction size should be capped since we haven't found a legitimate monetary reason for uncapped tx size. Bitcoin should gradually lock down the base layer consensus rules now that Bitcoin has stable L2 protocols that developers can build L3 apps on top of. The internet has lasted a very long time with very few base layer protocols. The application layer is the wild west where innovation and use cases should be developed. In protocol development, you should always keep the scope of lower layer protocols pure and stable so that upper layers may flourish. Good luck with the BIP!
5
4
40
2,668
Aiden 🏴 retweeted
It seems we might be having different conversations. Our point is that nodes have the ability to add, remove or adjust filters, any filters, period. This is an irrevocable, indisputable fact. The nature, importanance, or efficacy of any given filter is irrelevant, so debating why some filters exist and other do not is a waste of time. What is needed is a strategy for policy that maximizes the intersection of user safety, user choice, and network benefits. Aiden's BIP is the best idea out there, and that's what should be debated. High level strategy for policy on a go-forward basis, not debating specific filters and their defaults. It's time the community move on from that and get back to work.
3
1
31
1,893
**Is your feature request related to a problem? Please describe.** The integers are not closed under division, but we need to divide them anyway, which is a pain point **Describe the solution you'd like** Make all of the integers even, so we can divide them more easily
3
1
19
1,838
Immigration restrictions are affirmative action for natives with the same counterproductive consequences.
12
15
2
67
12,321
Thinking about the completely fake bomb/drug/criminal detection device that was sold for up to $8000 per unit to police and militaries the world over and was quite literally a useless piece of plastic full of strangely deliberately placed dead ants
175
3,744
330
24,029
Systemd. PulseAudio. Kernel DBus. All created by the same person. He's like the Fabrice Bellard of software people hate
12
41
7
686
HELLO EVERYONE I HAVE A DEEPLY ABSURD CURSED COMPUTER STORY
6
25
5
267
I get constant messages like this: "It's inevitable. Don't want to be tracked, don't use technology" Most people feel helpless when it comes to digital surveillance, but they're more empowered than they realize. Surveillance is absolutely not inevitable. Stop normalizing it.
21
42
4
180
concealed weapons permits should only be issued to people who can demonstrate a good reason to have one, such as having 200 spare iPads, access to a good box at NHL games, campaign donations, or just thinking it’s a “cool thing to have”
2
5
3
39
Aiden 🏴 retweeted
We're hiring @start9labs: 1. Frontend Dev: JS/TS experience (XP) required, Angular/RXJS XP preferred 2. Systems Dev: Rust XP required, Linux internals XP preferred 3. Product Designer: Figma/Sketch/equivalent XP required, HTML/CSS XP preferred DM me if interested
3
28
2
45