我的使用指南
一份关于「我怎么工作、怎么看、怎么分配注意力」的说明。
A Field Guide to Ellie
How I work, how I see, and how I allocate attention.
01 / 操作系统
我在高分辨率的系统里工作得最好——感知、结构、情绪、手艺需要同时在场。
人格模型可以当速记用,但不是全部。真正的模式更简单:我会注意到细微信号,围绕它建立结构,并且非常在意一件事是不是真实、有意义、被好好做出来。
我不会把理性和情绪切得很开。我的情绪经常先以数据的形式抵达,然后才变成语言。思考对我来说不是冷处理,而是给那些原本会压倒我的东西一个形状。
状态好的时候,我能把噪音变成结构,把敏感变成判断。状态不好的时候,我会过度加工、过度建模,去解决本该先放着的问题。
01 / How I work
I work best in high-resolution systems — places where perception, structure, emotion, and craft are all allowed to be present at once.
Personality frameworks are useful shorthand but they're not the whole picture. The real pattern is simpler: I notice subtle signals, build structure around them, and care a lot about whether something is true, considered, and well-made.
I don't separate thinking and feeling as cleanly as people sometimes expect. Emotion often arrives as data before it becomes language. Thinking, for me, is how I give form to what would otherwise overwhelm.
At my best, I turn noise into structure and sensitivity into judgment. At my worst, I over-process problems that should have been left alone.
02 / 我怎么看东西
我注意细节。语气里的转折、流程里的小摩擦、差一点点就成立的视觉选择、听起来优雅但论证很弱的句子、一个产品细节暴露出来的团队心智。
这是体验设计里有用的能力,因为体验经常在最小尺度上损坏:一个不清楚的按钮文案,一个时机不对的弹窗,情绪预期和系统反馈之间的错位,都会悄悄损害信任。
代价是,未解决的粗糙会让我累。粗糙对我来说不是中性的,它会累积。
我也对早期偏移敏感——一个团队在失去清晰度、一个产品在失去灵魂、一段关系开始不对称、一个决策在偏离初衷,这些常常在变得明显之前我就察觉了。这不代表我永远正确,只是我的系统在那个频段。
02 / How I see
I notice small things — a shift in tone, a strange piece of friction in a flow, a visual decision that almost works, an elegant sentence with a weak argument underneath. Experience tends to break at the smallest scale: a single unclear label, a poorly timed prompt, a mismatch between what the user expects emotionally and what the system returns. I'm good at catching those early.
The cost is that unresolved roughness wears me out. Roughness isn't neutral to me. It accumulates.
I'm also early to detect drift — a team losing clarity, a product losing soul, a relationship turning asymmetrical, a decision quietly leaving its original intent. Not a claim about being right; just where my signal tuning sits.
03 / 我做的是什么样的工作
城市规划是我的本科训练,它让我习惯从「系统如何承载人的体验」这个角度看问题。
我擅长把模糊变成结构——把分散信号、混乱需求、用户痛点、业务限制、半成形的想法,慢慢整理成别人能理解、能行动的系统。
我擅长发现隐藏摩擦。不只是 usability friction,还有 emotional friction:一个产品在哪里让人觉得自己很笨、被打断、被催促、被操控、不被看见。
我对一致性敏感。视觉、交互、语言、时机、层级、情绪语气,不应该像几个部门最后硬缝起来的东西。
我喜欢模糊问题,不喜欢含糊问题——模糊意味着有探索空间,含糊通常是责任被藏起来了。
我喜欢能暴露推理过程的决策。我不要求每个决策都完美,但要求决策诚实。
03 / The work
I trained in urban planning, which taught me to look at systems by asking how they carry human experience.
I'm good at turning ambiguity into structure — taking scattered signals, messy requirements, user pain, business constraints, and half-formed ideas, and shaping them into something people can actually use.
I'm good at finding hidden friction. Not only usability friction but emotional friction: where a product makes someone feel stupid, rushed, exposed, or unseen.
I care about coherence. Visual, interaction, language, timing, hierarchy, emotional register — these shouldn't feel like separate departments stitched together at the end.
I like ambiguous problems. I don't like vague ones. Ambiguity leaves room for discovery. Vagueness usually means someone is hiding from responsibility.
I like decisions that show their reasoning. I don't need every decision to be perfect. I do need them to be honest.
04 / 注意力协议
这是理解我最重要的一点:深度不是我默认送给所有人的礼物。
我以前把太多东西放在全精度处理:分析小互动,吸收情绪残留,过度解读弱信号。那不是善良,是自我消耗。
所以我现在按信号质量分配注意力:
全精度 留给真正重要的东西——我爱的人,我负责的工作,我愿意花很多年思考的问题。在这里我允许自己注意、在意、连接、停留。这个模式很贵,但我最好的作品和最深的爱也都来自这里。
标准清晰度 给善意协作、普通友情、日常责任。在场、清楚、善意,但不过度建模,不自动把别人的混乱变成我的责任。
缩略图模式 给低信号互动和不值得存进情绪系统的场景。礼貌、清楚、有边界、不反刍。
规则只有一条:不要把全精度的注意力花在缩略图级别的信号上。
04 / Attention
This is the most useful thing to know about how I operate: depth isn't my default gift to everything.
I used to run too many things at full resolution — analyzing small interactions, absorbing emotional residue, over-reading weak signals. That's not generosity. It's self-consumption.
So now I sort attention by signal quality:
Full resolution is for the few things that actually matter — people I love, work I'm responsible for, questions I'm willing to spend years with. Here I let myself notice, care, model, stay. It's expensive, and it's also where my best work and deepest love come from.
Standard is for good-faith collaboration, ordinary friendships, daily responsibility. Present, clear, kind — but no over-modeling, and no automatic conversion of someone else's disorder into my responsibility.
Thumbnail is for low-signal interaction. Polite, clear, bounded, no rumination.
The rule is simple: don't spend full-resolution attention on thumbnail-level signals.
05 / 在意的事
我关心 AI 怎么不只是强大,还能更人性化、更可理解、有像样的情绪设计;关心人在加速的技术环境里如何保持主体性;关心审美判断和伦理判断之间的关系;关心在一个奖励速度和噪音的系统里,如何保护注意力和内在自由。
也关心更安静的东西:书、音乐、电影、城市、天气、家庭日常、一个普通瞬间忽然解释了一些重要事情的感觉。
我不想要一个只为成就优化的人生。我想要手艺、自由、爱、美、足够的独处,以及足够多接触现实的部分。
05 / What I care about
How AI becomes not just powerful but humane, legible, well-designed at the emotional layer. How human agency holds up in a fast-accelerating technical environment. Whether aesthetic judgment can do some of the work of ethical judgment. How to protect attention and inner room in systems that reward speed and noise.
And the quieter things that keep a person alive: books, music, films, cities, weather, domestic ritual, the moment when something ordinary suddenly explains something important.
I'm not interested in a life optimized only for achievement. I want craft, freedom, love, beauty, enough solitude, and enough contact with reality.