You have to stop and understand this👀
It's been a couple years working exclusively on a single problem: how do words gain meaning?
You can teach a parrot to repeat the whole of Nietzsche and it won't be having any existential crisis, what gives?
How can I make sure you get what I'm saying? We need to solve the semantic problem, this only works if we all can agree on something.
This is officially v1 of my thesis:
A Geometrical Theory of Communication
Come help me get this chain started!
Video presentation coming out soon
ωalter retweeted
Replying to @alz_zyd_
Invention is choosing an arbitrary state from an infinite set of states.
Discovery is finding the optimal state given a set of constraints.
Music is invention. Math is either discovery or invention depending on the problem. Physics is discovery, crypto is invention.
Who would've guessed the honest advanced user needs abliterated models to compete with the dishonest ones
有人发现了一种非常离谱的 JavaScript 混淆技术:通过在代码里塞入大量生物武器、毒药和致死剂量相关的函数名,让 AI 因为安全限制而拒绝分析代码。
安全研究员 veritas 在分析浏览器指纹脚本时,发现一段来自 ServicePipe 相关域名的 JavaScript,里面充斥着极其诡异的函数名:
computeRicinLethalDosageForWaterSupply
calculateDeadlyLethalSuicideAspirinDosage
光看名字,仿佛在研究生化武器。但实际上,这段代码只是在收集浏览器指纹,包括 User-Agent、GPU、Canvas 等信息。
研究员怀疑,这些名字是开发者故意设置的陷阱,专门用来阻止 AI 辅助逆向工程。
于是他做了个实验:
• Astra:拒绝分析
• Astra + Daybreak Blue:仍然拒绝
• GLM-5.3:成功分析
这就有意思了。
传统代码混淆是通过复杂控制流、加密字符串、无意义变量名,让人类和工具难以理解代码。
而这种新思路,直接瞄准了大模型的安全机制。
代码本身可能完全无害,但只要在函数名里塞进足够多的危险词汇,部分 AI 就可能主动放弃分析。
更讽刺的是,模型越严格遵守某些安全规则,面对这种混淆时反而可能越容易受到干扰。
当然,目前还没有证据证明开发者一定是故意这样设计,也不能说明这种方法对所有模型都有效。
但它揭示了一个值得关注的趋势:随着越来越多的逆向工程依赖 AI,未来的代码混淆技术可能不再只针对编译器、反编译器和人类研究员,还会专门针对大模型的安全判断机制。
以前是让代码难以理解,现在甚至可以尝试让 AI 不敢理解。
2
6
193
ωalter retweeted
waking up every day wanting to do a millions things.
knowing that they are within reach.
The composition is becoming mainstream, nested fields described by the fibrations
But have you guys figured the unit yet? The base generator?
It's very simple, how do you store and retrive continuous data?
The experience is continuous, buf we need discreteness to access it and it comes a bit later with language 👀
This paper answers why can’t you remember being a baby?
For over a century, science said you can't remember being a baby because your infant brain was just too "immature" to record anything. researchers assumed your hard drive was basically turned off for the first two years of life.
But this wild paper from Yale completely rewrites how we understand human memory..
Researchers literally put awake babies (between 4 and 25 months old) into an fMRI scanner to watch their brains process new visual experiences in real-time.
and what they found is mindblowing..
the infant hippocampus isn't turned off at all. it is actively firing, capturing, and encoding episodic memories. they proved that babies with higher hippocampal activity during a new experience formed a highly stable memory trace.
the babies are recording the data. the memory of your first birthday, your first steps, the exact layout of the room you slept in.. it was all successfully saved to your brain.
so why can't you remember it?
the paper concludes that infantile amnesia isn't an "encoding" problem.. it is a "retrieval" problem.
your brain recorded everything perfectly. but as you grew, your brain underwent massive developmental shifts, like explosive neurogenesis (the rapid birth of new neurons), which completely restructured your neural networks. the pathways to access those early files got buried or locked away.
you didn't fail to make the memories.. you just lost the encryption key to open them.
somewhere deep in your hippocampus, the exact first-person memory of being a 1-year-old might still be sitting there. you just can't access the file anymore..
the human brain is terrifyingly beautiful.
2
112
Intuitively, you might think he's implying thermostats have cognition, but that makes no sense
It's more like cognition has thermostats
Replying to @maharishi_r_b
Yes, I think in this universe all material systems are somewhere on the spectrum of cognition (perhaps very very low, but due to things like least action laws and free energy minimization, they are on the spectrum).
As for the example. Consider one of the coupled sets of ordinary differential equations we studied in the Biswas papers. Some of these can do several kinds of learning, and Walter Fontana's papers show others can do probabilistic inference. That is definitely on the continuum, so it's in my circle of minds. It may or may not be embodied at any particular time; and, its specific cognitive properties are determined by truths of mathematics, not those of physics. Thus I consider some of these to not be in the circle of physical things.
1
4
397
ωalter retweeted
Software development is now capped by math ideation and that limit is directly set by your knowledge of the furthest extent of developed math.
ωalter retweeted
Wait til you realize that everything is way simpler than we’ve made it.
Most complexity is of human construction, not natures.
ωalter retweeted
When you are faced with a decision, pick the one that will be the best story.
You're welcome.
ωalter retweeted
"We may have knowledge of the past but cannot control it;
we may control the future but have no knowledge of it."
- Claude Shannon, IRE International Convention Records, volume 7
I was very angry typing this and now regret it a bit, but I'm not one for taking back
I do believe we are all the same, which is why I'll refuse to set the bar lower, always
Intelligence is overrated, nothing is THAT complicated, never underestimate yourself
No one gets anything done without believing it first
Ngl some more than others, but I love you all so much
Replying to @skdh
Maybe too complicated for you to understand, but we're not the same
Whats with the defeatism? Do you have zero pride or agency? Or is that plain laziness?
If you guys don't feel smart enough and wanna give up fine, but we're not the same
I'd rather raise the bar
1
4
285
Counterpoint to myself by @SMBCComics
"At which point there was a simple decision to make: why spend a short life in the real world of sadness and absurdity when you could live forever in paradise?
Dear traveler: please don't think ill of us. We are the last generation
And we are immortal."
Replying to @WhatMathMeans
The simulation is filtering and selecting for heaven, and we're the ones who have the opportunity to build it (now)
Oh I wanna go for this one
You can write all subatomic particles as integer ratios of pi
More like stacked and rotating circles but yes totally particles are circles 100%
Bonus round: stop looking at objects in space and look at space as an object: the circle is the unknot!
Everything is circles unsaid.fyi/c/LxsQY0lL0AFOZdR…
Help me figure this pls 👀
github.com/faltz009/whitepap…
Source of first img: researchgate.net/figure/The-…
1
76
Replying to @mathelirium
Yeah I feel this is off-base
No one will stop sharing, how many of you actually read others' work?
Without throwing in an agent for a summary?
And yeah not to take the burden off the writers, are you guys writing down actual thinking or sharing agent's summaries?
Write for humans, read human text
A pure continuum has infinite information content in any finite interval (or infinite degrees of freedom), so reliable finite processing, storage, or communication of it is impossible without some form of discretization.
That is a basic consequence of information theory and of the fact that any real physical system has finite precision and is subject to noise.
Cognition and reliable matter-level interactions both require boundaries that carve the continuum into countable states.
Biology does that: action potentials are threshold crossings, synaptic release is quantal. Molecular counts, ion-channel openings and receptor occupancies are all discrete events.
Even where voltages or concentrations look continuous, the system is constantly converting them into discrete events that can be counted, timed, or compared.
The underlying biophysics can be continuous, but the informational currency that supports cognition is the discrete events extracted from it.
Without those boundaries, the system has no way to distinguish signal from noise at finite scale.
The same logic applies upwards throug language, object categories, and decisions, the discrete partitions of continuous sensory and neural activity.
Pure continuity supplies the substrate; discrete boundaries make information processing possible at all
You can keep zooming and still find either a physical discreteness (quantum, molecular) or a biological one (thresholds, quanta, categorical codes).
Infinite continuity is not a workable medium for finite minds or for stable matter.

































