The precision writing gap

I believe writing is thinking. And personal intellectual leverage.

In October 2024, Paul Graham described a future of “writes and write-nots”. With AI, its easier than ever to write.

But not necessarily think. If writing is indeed thinking then the future he describes is one of “think and think-nots”— very scary indeed.

The future will depend on more than just being able to write.

Those using AI and LLMs know there will be a further divide. A separation between who can write from those who can write with precision.

Concision is often lauded, but precision is power

Precision is the ability to express ideas exactly. It means choosing words that have no ambiguity. They must capture the intended meaning.

It's about semantic accuracy.

Conciseness is brevity: using fewer words.

While these qualities often complement each other, they're distinct.

A precise statement might require more words than a concise one. Consider:

Concise but imprecise: "The project went well."

Precise but longer: "The project delivered 15% above revenue targets while staying within its $50,000 budget."

No carelessness in your actions. No confusion in your words. No imprecision in your thoughts. — VIII. 51

Tricks I use to be more precise

• First, I try to question my assumptions: When I want to write "good" or "better," I ask "compared to what?" When I write "soon," ask "when exactly?" We can find the needed precision by testing default abstractions.
• Second, I prefer concrete details over abstractions. Instead of, "The company is growing fast," I’d write, "The company increased revenue 40% year-over-year from $1M to $1.4M."
• Third, I define terms. I explicitly state my meaning when I introduce important concepts. In work, especially remote, we should not assume shared understanding.

Consider cooking and reading a recipe book. In some cases a general recipe may say: “Add spices to taste”.

Some chefs have an instinct for cooking. For beginners, it's up for debate.

A precise recipe states: "1/4 teaspoon ground cumin, 1/2 teaspoon coriander, 1/8 teaspoon cayenne pepper.""

The benefit? Its replicable.

An example with a LLM

Let’s use Claude to demonstrate the difference between being general and precise.

(General) User Prompt: Write a paragraph about climate change

Response: Climate change represents one of humanity's most pressing challenges, marked by a steady rise in global temperatures due to increased greenhouse gas emissions from human activities. Since the Industrial Revolution, the burning of fossil fuels, deforestation, and industrial practices have released unprecedented levels of carbon dioxide and other heat-trapping gases into the atmosphere, leading to an average temperature increase of 1.1°C globally. This warming has triggered a cascade of environmental impacts, including rising sea levels threatening coastal communities, more frequent and intense extreme weather events, disrupted ecosystems, and shifts in agricultural productivity.

(Precise) User Prompt: Write a paragraph analyzing how a 2°C temperature increase by 2050 would affect corn yields in Iowa, using historical drought data from 1990-2020 as a baseline. Focus on economic impact for farms between 100-500 acres.

Response: Based on historical drought data from Iowa during 1990-2020, a 2°C temperature increase would likely reduce corn yields by 17-23% for mid-sized farms (100-500 acres). This projection accounts for the documented relationship between each 1°C rise and an 8-11% yield reduction in the U.S. Corn Belt. For the average 300-acre Iowa farm currently producing 195 bushels per acre, this translates to a decrease of 33-45 bushels per acre. At 2023's average corn price of $4.75/bushel, these farms would face annual revenue losses between $47,000-$64,000. While larger operations above 500 acres can often absorb such losses through scale and diversification, these mid-sized farms typically operate on thinner margins with net farm income averaging $75,000-$125,000 annually. The projected losses therefore represent 40-85% of their current profit margins, potentially forcing many to seek additional financing or consolidate with larger operations, accelerating the ongoing trend of farm consolidation in the region.

The value of precision

Precision forces clearer thinking about what you actually want to know. It helps the LLM understand the context and constraints. It produces more actionable insights.

In this context, precision becomes a heightened form of literacy. This is vital in AI interaction. The output quality depends on the input's precision.

The true divide will be more nuanced and important. It will be between those who can articulate their thoughts with precision and those who cannot.

AI amplifies both our strengths and weaknesses in thinking.

• Vague thoughts become vague outputs, multiplied
• Precise thoughts become precise insights, multiplied
• Poor questions yield poor answers, at scale
• Exact questions yield exact answers, at scale

In conclusion

We're not just facing a future of writes and write-nots, or thinks and think-nots.

In the future, the ability to think and communicate precisely will divide us.

The question isn't whether you can use AI; it's whether you can think precisely enough to make AI think precisely for you.

That future that’s coming? It’s here.

Precision is its currency.