Everyone on earth takes a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press? BE HONEST.
44%Red
56%Blue
312,448 votes • Final results 5
277
Zen retweeted
My dear front-end developers (and anyone who’s interested in the future of interfaces):
I have crawled through depths of hell to bring you, for the foreseeable years, one of the more important foundational pieces of UI engineering (if not in implementation then certainly at least in concept):
Fast, accurate and comprehensive userland text measurement algorithm in pure TypeScript, usable for laying out entire web pages without CSS, bypassing DOM measurements and reflow
Zen retweeted
AI agents are powerful — but they weren't trained on ICP.
ICP Skills gives them real knowledge — so they can build sovereign software on an onchain open cloud that's tamperproof, unstoppable, and handles digital assets and payments.
skills.internetcomputer.org
Zen retweeted
Excited to release @gepa_ai's optimize_anything: a universal API for optimizing any text parameter.
It consistently matches or outperforms domain-specific tools optimizing code, prompts, agent harnesses, cloud policies, even visuals!
If you can measure it, you can optimize it.
Zen retweeted
DFINITY Foundation and the Government of Pakistan have signed an MoU to establish sovereign cloud infrastructure, generative AI app development, and secure national communications, on the Internet Computer.
The collaboration includes:
— Pakistan Subnet
— Caffeine AI Licenses
— National Messenger Service
— DFINITY in Pakistan
Through this decision, Pakistan furthers its digital sovereignty and AI push in line with announcements made by the prime minister at Indus AI Week.
1/4 🧵
Zen retweeted
I didn’t see much Internet Computer discussion on Substack, so I decided to start publishing there.
There’s a serious investor/VC audience on the platform that feels underexposed to $ICP.
Would love feedback from the community on this piece
open.substack.com/pub/theswo…
Few thoughts, take with a grain of salt...
We don't have enough information on the blob storage network. What's the replication factor? What does the node spec look like? Who can become a node provider? How is long-term data integrity managed? What're the economic implications? This information should be available during these discussions.
I was opposed to increasing compute costs before Mission70 was announced. Currently, I'm cautiously in support of it. Costs will need to increase at some point and whether they change now, in 6 months or in 2 years - developers will need to adjust in anticipation of what those costs will be. I have no thoughts on the 5x figure - the goal is sustainability (assuming demand). Delays on this would be a courtesy to developers (completely fine with that), but we'd be kicking the can down the road. If we assume new enterprise interest, it would be preferable to have the cost model finalized before then.
Cloud engines sound cool but whether this vision will pan out remains to be seen.
Sounds like Caffeine will be developing tools for people to leverage locally which is good news. Also will support import/export...devs can create their apps with their own tools and workflows then import and deploy through Caffeine.
Fine with the dissolve delay changes and voting reward changes. TLDR: 2 weeks min, 2 years max - 8YG neurons get 10% boosted rewards until 2030. Rewards scale non-linearly.
Support capping the rewards pool. Support the modulation changes.
Support decreasing Gen-1 node rewards. No opinion right now on the smaller SEV subnets.
~50% of nodes are currently unassigned. The whitepaper doesn't touch on the impact of performance based node rewards and I'm hazy on those details but it does mention that 50% spare capacity is excessive which I agree with. I would support aggressive measures in relation to node providers generally. Gen-2 and Gen-3 node providers aren't affected by Mission70 to the same extent that investors and developers are. I don't have a good grasp on what measures would even be possible in this area.
teapawt.org/dominic_w/status/20190…
UPDATED Mission 70 white paper — proposing to 5X the cost of ICP's onchain compute as provided sovereign cloud will remain very competitive.
↗️↗️↗️ Caffeine & cloud engine demand
↗️↗️ cost of onchain compute
↙️ inflation (gov + nodes)
🔥 huge deflation
internetcomputer.org/whitepa…
UPDATED Mission 70 white paper — proposing to 5X the cost of ICP's onchain compute as provided sovereign cloud will remain very competitive.
↗️↗️↗️ Caffeine & cloud engine demand
↗️↗️ cost of onchain compute
↙️ inflation (gov + nodes)
🔥 huge deflation
internetcomputer.org/whitepa…
154
Caffeine has definitely improved. Here's a side-by-side between Kimi K2.5 agent mode and Caffeine. A detailed PRD was provided as the initial prompt - the goal was build a drum sequencer.
Kimi oneshot the app. With the exception of minor bugs the app works perfectly.
Caffeine produced similarly good results on the frontend but the bugs on the backend were more severe and I couldn't fix them before running out of credits.
Regardless, users should expect much better results with the new system. Model-agnostic / multi-agent / self-improving is the way to go and from looking at the new JSON artifacts being produced during development, they are optimizing in this direction and things will only get better and faster.
Zen retweeted
Following up on my last post about RWA infrastructure:
5 non-negotiables for a $16 trillion market:
1. Canonical Ledgers
One source of truth. Tamper-proof. Auditable. Final.
Settlement failures cost $3B annually (DTCC).
2. On-chain Legal Documentation
The deed to your tokenized building can't live on a hackable server.
Legal docs must be stored ON the blockchain, not linked to it.
3. Zero-gas Predictability
You can't tell BlackRock their $500M transaction might cost $50 or $5,000.
Enterprise needs fixed costs.
4. Multi-chain Reach
Assets need to move across chains without 47 bridges and 12 wrapped versions.
Issuers must retain control.
5. Sovereign Infrastructure
Not "decentralized" while hosted on AWS.
True sovereignty with regulatory clarity for compliant RWA issuance.
Most solutions deliver two or three. Almost none deliver all five.
Read my full take: teapawt.org/PierreSamaties/status/…
Zen retweeted
Here we go!
1st national subnet, more to come!
WCD, 2026
Super excited, @SwissSubnet 1st national subnet launched @ WCD, Davos
Deploy apps on sovereign Internet Computer/ICP territory keeping all data and computation in Switzerland, to meet emerging regulatory and sovereignty objectives, and more
Next, Swiss Subnet cloud engines 💪🏼
Zen retweeted
Major upgrades in the pipeline!
Agentic building will make Caffeine much more capable! Caffeine will understand you better, it will make less errors. It will handle larger codebases, and many times it will be faster.
Stay tuned.
Feb 2nd, Caffeine.ai will begin to roll out it's all-new fully agentic "automated tech team" (v2.0)
Big advances coming for self-writing cloud and ICP ☕💪
Zen retweeted
An important new ICP white paper. This is a big one, enjoy! #Mission70
"Mission 70 and Accelerating the Internet Computer Economy"
internetcomputer.org/whitepa…
Zen retweeted
#Mission70 paper is on its way as promised :)
Once that's shipped tomorrow, I'm going to announce an already imminent date for Internet Identity 2.0! Generally speaking, 2026 is going to involve a lot of big announcements, and a lot of impactful new things, so buckle up 💪
A new DFINITY white paper proposing how #Mission70 (to reduce ICP inflation 70% during 2026) can be won, shall be published next Wednesday, 14th Jan. Soon after we will create an NNS proposal. Forward ICP 🔥
What is Sector9?
Whitepaper: github.com/Neutrinomic/secto…
#ICP is top-tier #crypto infrastructure. You can’t build something like this anywhere else-not without permission to modify the L1, and not while remaining scalable, fast, and privacy-preserving.
Overview (simplified):
Confirmation that Anthropic is intentionally blocking OpenCode, and any other 3P harness, in a paranoid attempt to force devs into Claude Code. Terrible policy for a company built on training models on our code, our writing, our everything. Please change the terms, @DarioAmodei.
"when you feel like it" No, a terms violation? Do you think we shouldn't enforce our terms? If anyone wants to use Claude models, they simply use the API. The one thing we should've done better here is tell end users sooner, as they might not have known. teapawt.org/trq212/status/20096898…





















