Pinned Tweet
"A portfolio of just cash and crypto, if sized correctly, will allow one to stay retarded and alive longer than most others can stay solvent.
That’s my fortress of fucking solitude."
🤝
Why doesn't the fed throw lawmakers(congress/senate) under the bus and tell them to balance the budget?
Rates should only change during the worst of circumstances. If rates are considered for change every few months, we're doing it wrong.
11
super intelligence
Controlled by few or gifted to all of humanity?
No straddling the line on this.
14
Start reading tweet
Go to another app to reference info
Come back to the tweet... gone
Rage
Auto refreshing the feed has to stop.
12
Everyone is nervous about the financial shenanigans of the AI build out.
If it trips up. Be prepared to bid tf out of it.
The US gubmnt will not allow it to fail.
Each major Corp has a blank check to let it fly.
14
Tech has always won in the long run.
Dot com bubble was just a little blip.
The world’s most accomplished tech founders and business leaders are betting hundreds of billions on AI with extraordinary conviction. Meanwhile much of the investing world thinks they’re making a massive mistake. The irony is hard to ignore. An investment industry where very few have historically managed to beat a passive index is now questioning the judgment of the people building the technology and running the businesses that will put it to work. Maybe the investors are right. Maybe the builders are. But I can’t remember a time when the stakes were this high and the two sides saw the future so differently.
18
This guy gets it. This has been the standard for my teams for over a decade.... led by me. It was always an uphill climb with new dev teams, but they eventually saw the light.
我们之前招的10年经验DevOps,也会自动化测试。花了2个月,给我们公司一个只有二十多万行的webapp产品,生成了5000多个单元测试。只查出来几不痛不痒的bug。
因为单元测试依赖于实现,每次开发们改代码,都要花几十分钟,删除老的单元测试,重新生成新的单元测试,怨声载道。
这个DevOps离职以后,我不让老板再招人,我自己接手这个自动化测试。我定了一个原则,测试用例只测用户能感知到的东西,绝对不基于任何内部实现写测试。我还写了一个通用测试技能,只让GPT生成150个测试用例,就抓到了15个bug。
我的结论是:一般的产品不要再写单元测试。多人长时间vibe coding出来的东西,可能核心功能能用,但是用户可见的边边角角的bug率非常高,如果没有专门的工作流去测试,很难发现。
9
💯
shower thought: public opposition to ai will flip 180 once physical ai reaches the masses.
for most ppl digital ai is either just a nice-to-have or an existential threat to their job. physical ai actually removes real stress from daily life.
case in point, tesla fsd has already had a bigger impact on mine than every chatbot combined (even tho the chatbots have been insanely valuable to me)
14
So, for those of you who missed it, Illinois just dropped its draft rules for their no good, very bad Digital Asset Tax and, y’all, it’s a mess, especially for DeFi.
The proposed rules would treat a DEX (not an interface, the protocol itself) that charges a “protocol fee” as a “broker” that must register with the state, collect a 0.2% tax on the full value of every swap by an Illinois user, and keep each user’s address, IP, and W-9. Effectively mandated KYC.
It also requires brokers to keep customers’ private keys… I don’t even know where to begin on that one.
Essentially a worse version of the repealed IRS broker rule.
Thankfully @BlockchainAssn, @crypto_council, and the @DigitalChamber have sued the state, but we need a full court press on this from a political and policy perspective.
It’s my personal #1 priority, because I doubt Illinois will be the last to try this.
This has always been the case.
If you needed unit tests, your developers were... not good.
Unit tests were for an era when the 50 percentile programmer could barely one shot a LeetCode medium.
You don’t need unit tests 99 times out of 100 if you can now one shot LeetCode hards with a well-defined spec.
44
Unit tests were never a thing(that mattered)
IMO the only tests you should have in the age of coding agents are, in order of priority:
1. Full E2E tests. Nothing mocked out at all. Can even be something that runs in prod on test accounts via Playwright or equivalent.
2. Integration tests. Agents make more mistakes as things bleed between data/API boundaries where schemas can drift.
3. Golden tests. This helps ground the code with real examples of data, and can be used as regressions against edge cases.
Unit tests are bloat 99% of the time if you’re using a frontier model since they’re smart enough now to get the implementation (and subsequent iterations) right on the first shot, so they only add bloat at this point.
20
It's 2026...
This is no longer something you should concern yourself with. Separate it. Keep it. Either option is a few minutes from a PR..
Move on.
The DRY (Don't Repeat Yourself) rule causes more architectural damage than code duplication ever will. Engineers force two completely different business modules to share a single abstraction just because they look similar today. When those modules inevitably evolve in different directions, that shared abstraction turns into a tangled mess of conditional logic.
17














