{"chain":[{"channel":"cities","content":" <<< Lidgerwood is a small city located in Richland County, North Dakota, with a population of approximately 591 as of 2024 . Founded in 1886, it is part of the Wahpeton, ND\u2013MN Micropolitan Statistical Area . >>>\r\n\r\n----\r\n\r\nONE. The \u5730\u7ea7\u5e02. ( my first guess was \u81ea\u52a9\u9910). China is almost entirely divided into \"Prefecture-level cities\". Or, second-level administrative divisions, named after cities.\r\n\r\nThese are, mostly, equivalent to a US county. And have about the same irregularity. ( Los Angeles County is huge, and much more than LA City. Baltimore City is only part of Baltimore Metro. Hardin County, Iowa is a roughly 24-mile square of Iowa, with no regionally-prominent urban areas.)\r\n\r\n--MORE--\r\n\r\nTWO. Atacama request-log.\r\n\r\n# Implement the << more >> tag.\r\n# Fix the issues with nested tags in the RSS feed.\r\n# Maybe have the *machine* re-write everything, again.\r\n\r\n----\r\n\r\nThis is, ultimately, *iterative* development. Take version 3915b97dd3c518af from January, and turn it into d88a12b9ab43d69f from April.\r\n\r\nThe Github revision history will reveal a series of iterative changes. ( one can *imagine* other ways of describing those changes, it is a much more difficult task to describe the techniques of description.)\r\n\r\n----\r\n\r\nPerhaps \"checksum management\" features are part of this.\r\n\r\n this might be the project I thought I needed 2 BTC for back in 2017.\r\n I no longer think Bitcoin Ownership is necessary.\r\n\r\n----\r\n\r\nThe topic of a [[Swadesh list]] came up again. It is a list of words found in rare languages. Used as a type of \"linguistic universal\" reference, or as a way to assemble similarity-trees of languages.\r\n\r\n the vocabulary is very heavy on body-parts and animals, and very light on Critical Race Theory.","created_at":"2025-05-05T04:53:49.048640","id":466,"is_target":false,"parent_id":null,"processed_content":"

Lidgerwood is a small city located in Richland County, North Dakota, with a population of approximately 591 as of 2024 . Founded in 1886, it is part of the Wahpeton, ND\u2013MN Micropolitan Statistical Area .

\r

ONE. The \u5730\u7ea7\u5e02. \n \ud83d\udd25\n ( my first guess was \u81ea\u52a9\u9910)\n . China is almost entirely divided into \"Prefecture-level cities\". Or, second-level administrative divisions, named after cities.\r

\n

These are, mostly, equivalent to a US county. And have about the same irregularity. \n \u2699\ufe0f\n ( Los Angeles County is huge, and much more than LA City. Baltimore City is only part of Baltimore Metro. Hardin County, Iowa is a roughly 24-mile square of Iowa, with no regionally-prominent urban areas.)\n \r

Read More ...

TWO. Atacama request-log.\r

\n
    \n
  • Implement the more tag.\r
    • \n
  • Fix the issues with nested tags in the RSS feed.\r
    • \n
  • Maybe have the machine re-write everything, again.\r
    • \n

This is, ultimately, iterative development. Take version 3915b97dd3c518af from January, and turn it into d88a12b9ab43d69f from April.\r

\n

The Github revision history will reveal a series of iterative changes. \n \ud83d\udd25\n ( one can imagine other ways of describing those changes, it is a much more difficult task to describe the techniques of description.)\n \r

Perhaps \"checksum management\" features are part of this.\r

\n

\n \ud83d\udd25\n this might be the project I thought I needed 2 BTC for back in 2017.\r\n

\n

\n \ud83d\udca1\n I no longer think Bitcoin Ownership is necessary.\r\n

The topic of a Swadesh list came up again. It is a list of words found in rare languages. Used as a type of \"linguistic universal\" reference, or as a way to assemble similarity-trees of languages.\r

\n

\n \ud83d\udd25\n the vocabulary is very heavy on body-parts and animals, and very light on Critical Race Theory.\n

","subject":"lidgerwood"},{"channel":"cities","content":"what does the << LLM-powered accounting system >> look like?\r\n\r\n# you email it your expenses\r\n# it categorizes them\r\n# it << balances your checkbook >>\r\n# it texts you with questions\r\n\r\n----\r\n\r\nthere are obvious << extensions >>:\r\n\r\n# A product that gives budgeting advice.\r\n# A product that gives financial advice.\r\n# A tax product.\r\n\r\n one of the problems is that humans, predictably, dislike sound advice. they prefer the advice that says they can continue doing the problematic thing, and the inevitable consequences will disappear by magic.\r\n\r\n----\r\n\r\nthe \"data archiving\" is a solved problem.\r\n\r\nall you need is:\r\n> a database\r\n> a blobstore for the original artifacts (receipts, emails, monthly snapshots of the bank account)\r\n> a machine-readable summary of the artifacts\r\n\r\n----\r\n\r\nas far as << asset management >>: you cannot reasonably avoid it. ( do you keep petty-cash in a USD-denominated account? Gold? SPY? ( S&P 500 Exchange-Traded Fund) )\r\n\r\nas far as << asset custody >>: this is not a current plan.\r\n\r\n----\r\n\r\nthe various \"views\" will surely include a << monthly spend >> view, as well as a << recent transactions >> view.","created_at":"2025-05-05T18:27:54.854914","id":467,"is_target":false,"parent_id":466,"processed_content":"

what does the LLM-powered accounting system look like?\r

\n
    \n
  • you email it your expenses\r
    • \n
  • it categorizes them\r
    • \n
  • it balances your checkbook\r
    • \n
  • it texts you with questions\r
    • \n

there are obvious extensions:\r

\n
    \n
  • A product that gives budgeting advice.\r
    • \n
  • A product that gives financial advice.\r
    • \n
  • A tax product.\r
    • \n
\n

\n \ud83d\udd25\n one of the problems is that humans, predictably, dislike sound advice. they prefer the advice that says they can continue doing the problematic thing, and the inevitable consequences will disappear by magic.\r\n

the \"data archiving\" is a solved problem.\r

\n

all you need is:\r

\n
    \n
  • a database\r
    • \n
  • a blobstore for the original artifacts (receipts, emails, monthly snapshots of the bank account)\r
    • \n
  • a machine-readable summary of the artifacts\r
    • \n

as far as asset management: you cannot reasonably avoid it. \n \ud83d\udca1\n ( do you keep petty-cash in a USD-denominated account? Gold? SPY? \n \u2699\ufe0f\n ( S&P 500 Exchange-Traded Fund)\n )\n \r

\n

as far as asset custody: this is not a current plan.\r

the various \"views\" will surely include a monthly spend view, as well as a recent transactions view.

","subject":"lidgerwood, part 2: account-a-rama"},{"channel":"cities","content":"The base magic is that, when a \"checksum\" is included, there is a link to the referenced content.\r\n\r\nFor Git repositories, this might need to include a full path to the repo.\r\n\r\nBut it is (or at least should be) a way of asserting a datetime and content.\r\n\r\n----\r\n\r\nThere are two basic functions needed for *magic*.\r\n\r\nOne is << delayed-release >>. The checksum is released now, the content is released in 1 year.\r\n\r\nThe other is << expiration >>. The checksum is released now, the ability to \"forge\" a checksum is released in 1 year. Now, one can have arbitrary content with the same checksum.\r\n\r\n----\r\n\r\n as \u5c0f\u9f99\u8bed is not available, we will use JSON.\r\n well, actually ... this all seems needlessly complicated. because of trying to get one system to do two very different things.\r\n\r\n----\r\n\r\nstep 1 is << add support for Github checksums >> to earlyversion.\r\n\r\nstep 2 is << some type of blob storage with metadata >>.\r\n\r\nonce i stop needing to tie them together, the need for *magic* dissipates.","created_at":"2025-05-07T15:46:02.313760","id":469,"is_target":true,"parent_id":467,"processed_content":"

The base magic is that, when a \"checksum\" is included, there is a link to the referenced content.\r

\n

For Git repositories, this might need to include a full path to the repo.\r

\n

But it is (or at least should be) a way of asserting a datetime and content.\r

There are two basic functions needed for magic.\r

\n

One is delayed-release. The checksum is released now, the content is released in 1 year.\r

\n

The other is expiration. The checksum is released now, the ability to \"forge\" a checksum is released in 1 year. Now, one can have arbitrary content with the same checksum.\r

\n \ud83d\udd25\n as \u5c0f\u9f99\u8bed is not available, we will use JSON.\r\n

\n

\n \u2694\ufe0f\n well, actually ... this all seems needlessly complicated. because of trying to get one system to do two very different things.\r\n

step 1 is add support for Github checksums to earlyversion.\r

\n

step 2 is some type of blob storage with metadata.\r

\n

once i stop needing to tie them together, the need for magic dissipates.

","subject":"lidgerwood, part 3: checksums and magic"},{"channel":"cities","content":"Through sheer desperation, I have stumbled onto the << concatenate all 20 source files into a single file >> approach of preparing the key files in a Git repo for Gemini. Or ChatGPT.\r\n\r\n none of them can accept \"here is the Github repo\". There are two clear reasons why. One is the problems with << partnerships >>, and in general that the LLM companies would rather let other companies build the coding tool. The second is that they don't have a smart way of limiting the data used per-request; the naive approach of reading all the data doesn't scale.\r\n\r\n----\r\n\r\nUnfortunately, both ChatGPT and Gemini are giving unsatisfactory results. ( well, actually, the results Gemini gives are fine ... the free usage is just so minimal that it is not useful)\r\n\r\n----\r\n\r\n... I'm not thinking fast enough to make << coding tasks >> a worthwhile use of time today. Time to do something else.","created_at":"2025-05-07T17:03:42.944882","id":470,"is_target":false,"parent_id":469,"processed_content":"

Through sheer desperation, I have stumbled onto the concatenate all 20 source files into a single file approach of preparing the key files in a Git repo for Gemini. Or ChatGPT.\r

\n

\n \ud83d\udca1\n none of them can accept \"here is the Github repo\". There are two clear reasons why. One is the problems with partnerships, and in general that the LLM companies would rather let other companies build the coding tool. The second is that they don't have a smart way of limiting the data used per-request; the naive approach of reading all the data doesn't scale.\r\n

Unfortunately, both ChatGPT and Gemini are giving unsatisfactory results. \n \u2694\ufe0f\n ( well, actually, the results Gemini gives are fine ... the free usage is just so minimal that it is not useful)\n \r

... I'm not thinking fast enough to make coding tasks a worthwhile use of time today. Time to do something else.

","subject":"lidgerwood, part 4: feeding the machine"}]}