prev Weekly note 2025-02-15~2025-03-01
It had been two weeks in which the rhythm of the week had been destroyed.
Please think deeply and advise from the following perspectives
--
.
1.### Respond to urgent bugs/defects . - T0036: Repair Scrapbox automatic English-to-English translation system (bug-fix, urgent). - It is also labeled "urgent" in the bug-fixing system, and the first priority is to resolve this to remove the bottleneck of "translation stoppage".
2.### Measures to deal with increased information channels . - T0116: Consideration of measures to cope with increased information channels. - Creating a mechanism to organize information that is scattered across multiple channels such as Signal/Slack/ChatGPT. High priority as it is expected to have a significant impact on reducing daily confusion.
3.### Operational improvements that lead to work efficiency and stress reduction . - T0098: Verification of Notepad + Night AI input flow. - A system in which the day's memos are submitted to the AI at night for tasking is effective in reducing omissions and confusion.
4.### Tasks that are highly effective in strengthening infrastructure/infrastructure . - T0099: Developing Task Management Scripts Using ChatGPT 4.0 API. - T0052: Time constraint handling functionality in show_next_action.py. - T0035: Implementing a Task Persistence Assurance System. - These are the foundations for more automation and more reliable task management. The priority is somewhat higher, but it is realistic to wait until after the above measures against confusion have been settled down.
--
.
--
.
1.### Start the highest priority task . - T0036 was started first, and the translation stoppage due to a bug was resolved. - In parallel, T0116 (information channel organization) will be studied and tried to operate together with the nightly summary flow (T0098).
2.### Specify tasks and clarify deadlines and time required . - If the task is vague, write down the "objective," "completion conditions," and "next step to be taken" to make it easier to make a decision to start the task. - If possible, set up even a rough resource estimate (time required and timing of implementation) and decide in advance where it will be executed on a weekly basis.
3.### Priority implementation of small automation tools . - Rather than implementing large scale functions, implement the parts that are easy to see the effects first, such as "Memo → summarized tasks at night," and test them out in operation. If the results are visible, motivation will easily increase.
4.### Make it a habit to reflect and organize once a week . - Fix "when and where to look back" every week, and be sure to check progress and backlogged tasks. - If a task is too vague to move, make an immediate decision to either "materialize the content OR reduce the scope OR put it on hold" and do not stagnate the backlog.
--
.
1.### T0036 (Scrapbox Automatic English Translation System Repair) ] - Start with highest priority. Proceed in three steps: isolate the cause, make tentative corrections, and test, and restore the translation as soon as possible.
2.### T0116 (Measures to cope with increased information channels) . - First identify "what is causing the confusion". - Examples: "too many Slack channels", "neglecting Signal logs", etc. - Next, tentatively decide "how to make it easier to summarize" and start [* with small experiments. - For example: "The nightly summary flow will be paper memo-centric, with AI assisting only where necessary.
3.### T0098 (night AI input flow) + paper memo operation . - Zap and record in paper notes from noon to evening → Ask AI to assist in organizing during evening session → Next morning, be ready to start on paper basis. - A good balance of minimizing AI dependence while leaving only organizing to the AI.
4.### T0099 (replacement for 4.0 to o1) . - If local execution, the timeout problem is low, so make MVP assuming o1. - First, design a simple API call part, check its operation, and then extend the functionality.
5.### T0052 (show_next_action.py time constraint support) ] - Currently low priority. "Simple instructions at the prompt" is sufficient, so it should be improved only occasionally.
6.### T0035 (Task Persistence Assurance) . - As long as there is no problem with Git management, it can be put off until later. - In preparation for in case of Web service, it would be smooth later if you only summarize requirements such as "backup frequency/locking function/differential management" etc.
Other "things that concern me" that I can think of
1.### Repair Scrapbox automatic English translation system (highest priority) . - T0036 applicable. Restore stopped functions as soon as possible.
2.### Prepare materials for 3/9 LT presentation . - Since the date is imminent Create slide skeleton by the end of the week -> Brush up at the end of the week - just before the end of the week.
3.### Plurality Tokyo Namerakaigi's video editing . - There is no deadline, but it is better to get started before the presentation loses its freshness; it is OK to start in the week after the 3/9 presentation when there is a little more time.
4.### README maintenance of public opinion map → OSS release . - No deadline. There is enough time to work on it after the English translation system is settled.
5.### Create a mechanism to allow Devin to do research . - Larger projects. It would be more efficient to organize the specific goals and means (e.g., how to extend Devin) before moving forward.
6.### Read Gmail to Gemini . - This is also a part of "measures to increase information channels," but the priority can be put on the back burner while assessing implementation difficulty and risk.
--
In this way, you can easily start working on long-term tasks while meeting immediate deadlines.
I believe the T0036 Automatic English Translation System Repair is an error that says that duplicate titles cannot be imported. This is on track to be resolved. Tomorrow is Monday and I can use the whole day for development. I think it would be delicate to use it only for 3/9LT preparation. I think it would be better to do something that requires concentration first tomorrow, but what do you think? It might also be a good idea to improve the task management system itself. Please look at the task backlog I give below and advise.
1.### First, complete T0036 (Scrapbox Automatic English Translation System Repair) . - Since the cause has already been mostly identified, I recommend a quick cleanup first thing in the morning to give you a "one done" boost in terms of mood.
2.### Start tasks for improvement around the task management system . - Example: - T0012 "AI Instance Role Definition Documentation" - Systematic documentation is maintained based on T0011 (completed). This is an easy task to concentrate on. - T0040 "Develop Task ID Safety Change Script" - Because of the dependencies and merge history involved, it takes a lot of concentration to review and implement the specification. - T0099 "Developing Task Management Scripts Using o1 (Local Execution)" - This is an area where we can sit back and create rather well as we proceed with API replacement with local execution.
3.### Instead of devoting a "full day" to preparing materials for the 3/9LT tomorrow, we will do it at a lighter time, such as between sessions or in the evening . - It is easier to stay productive and less likely to get bored if you do intensive code and design tasks in the morning and afternoon, and put together a small amount of material at night.
--
What we did this week
English translation system restoration
Add UI for adding tasks to AI task management system (for people to add tasks)
Added UI for review when AI suggests additional task updates to AI task management system
Plurality Tokyo Namerakaigi on YouTube
Unexplored Night LT
Take the parliamentary proceedings with the National Diet Library API and visualize them with the Yasuno team's new system.
Install lifelogging app on LINKLET and bring it to unexplored conference/unexplored night
This page is auto-translated from /nishio/週記2025-03-01~2025-03-08 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.