It is now the evening of February 5, 2026.
After reading C1's work log on his blog, I wanted to understand the details of what he's been busy with lately. It feels like he doesn't update very frequently, but he also mentioned his reasons for not recording daily:
The past few days have been quite packed, so I didn't record anything. However, stopping the recording is not an option, so I'm picking it up now and recording everything together. Besides being busy, why didn't I record?
- Because I was afraid that recording would require sitting down and spending over 30 minutes specifically to record a day. This is actually due to some fear and burden regarding recording daily life, which is not advisable.
- Usually, I only want to start recording a day after it has truly ended. But upon careful thought, this is somewhat anti-human because I now basically go to bed as soon as it's time to sleep, not because everything I wanted to do is truly finished (does such a time even exist?). This leads to not recording when there is time, and when it's really time to record, I have to quickly get into bed, compounded by issue 1.
The combination of these two leads to an accumulation over time.
My comments are:
Fragment Writing is Better Than Complete Writing
Writing fragmented records instead of complete records is a win-win situation for both AI and humans. There's no need to wait until the "complete day" ends to record. Write whatever comes to mind, and a day doesn't necessarily have to result in only one log entry. For example, I might post several log entries daily, all fragmented in content, though each entry's theme isn't too scattered. This is a more AI-friendly way for compression (summarization). AI can easily extract the core information from fragmented content and help you integrate it into more systematic knowledge. However, if you write a single log entry with overly cluttered content, AI will have a harder time extracting the key points.
Pre-Writing is Also Important
During the Thinking phase, AI engages in pre-writing, and humans should do the same. Pre-writing and post-writing are complementary. Pre-writing helps clarify thoughts, while post-writing aids in summarizing and reflecting. Neither should be omitted.
With pre-writing, even if the planned task isn't completed, you at least have a record of the Thinking process, which can help you pick it up better next time and provide key information for AI design. Sometimes, I directly throw several of my blog posts to an AI Agent, or even give it the entire blog, letting it further plan according to my train of thought. It can do quite well now.
Pre-writing doesn't require much extra effort. With the current level of inline guess completion functionality, you only need to write a few words, and AI can help you complete it into a full paragraph of thought. You just need to give it a general direction, and it can help you write out the thought content. This way, you can easily accomplish pre-writing.
Pre-writing is simply a huge win.
About Cross-Platform Comments
Speaking of which, why must my comments on his log be saved on his blog?
I'll just save them on my own blog. I only need to reference his content to comment; I don't really care whether visitors to his blog can see my comments. Whether he sees them, or other third parties see them, that's a matter of interconnectivity for internet products to solve. Comments, in essence, are "reference + commentary," and the referenced content can be any public content. Comments don't necessarily have to be saved on the platform of the person being commented on.
So, what if the content I commented on is revoked by him? It doesn't matter, because I don't need to prove that he actually said something. I just need to record my perspective on something. The referenced content is only to let readers understand the context of my comment. If the background excerpt is complete, who cares whether the original text still exists?
Preserving the content of the person being commented on is a problem for internet archives to solve. And I, as a commenter, indirectly preserve the content of the person being commented on as well.
The data ownership of comments lies with the commenter, not the person being commented on. This is the logic comments should have.
However, cross-platform comments, as an interesting technology, can actually be implemented. This is similar to backlinks. It just requires jointly rendering content from both sides. If we both use CZON to build, we can pull each other's meta.json files and then render the comment content within our own CZON systems. This way, comments can be saved cross-platform.
But I'm curious, is there a content system that cannot be revoked? That would probably require a publicly auditable blockchain system, right? As an author, when would you absolutely need to write to a non-revocable content system?
