Oceans Risk, Empires Fall
Title inspiration: You’ll Be Back
Joan Westernberg uses the failure of empires as a metaphor for why current predictions about AI are likely not to be completely realized. She equates empires with corporations and says this:
We haven’t seen the first great AI collapse. We haven’t seen a foundation model company make the BlackBerry mistake or the Nokia mistake, or the Roman mistake, or the Ottoman mistake or reach their Bunker-in-Berlin mistake. But we will, and we’ll see it multiple times, because these mistakes = features of power concentration. The hubris that makes a company or an empire dominant in one era is frequently the quality that blinds it to the next one. If you could ask Lazaridis in 2006, or a British colonial administrator in 1900, whether their model of the world was permanent, each would’ve given you a very convincing explanation for why it in fact was.
The interesting thing is that while she is applying the fall of empires to AI, she could actually be describing what is happening right now in the United States. We are living in a period of time in which the fall of the U.S. empire is accelerating. I think AI will be a contributor to the fall of the U.S. empire because employment is a pillar of the empire that has been built using economic might.
Those of us who have been raised on the privileges of the U.S. empire are told that the source of our pain is the “other,” however that is defined (race, nationality, gender) by each one of us all branded by the word terrorism. The purpose is to self preservation for the people in power who created and maintained the U.S. empire for their personal benefit, but as Joan points out history shows they too will ultimately fail.
That anxiety you are feeling is being caused by the stones of the U.S. empire starting to fall around you. All you have known is the life the empire has provided to you and you know your life is going to change. We know this in our bones because change is a cosmic truth.
Da da da dat da dat da da da da ya da
I wish that the micro.blog editor had a “source code” mode. Yesterday I uploaded a screenshot from my phone and after viewing in in a desktop web browser I wanted to reduce the size by half. I went to edit the post but found no way to actually see the img html tag to change the width and height. I had to remove the picture, go in to Uploads, copy the HTML, then go back to the post and add the HTML and then change the width and height.
It is absolutely gorgeous outside right now. Have a great day!
I spent the afternoon working on and testing Scott Hanson’s dockerized version of Feedland. I’ve written notes but I don’t have the time now to edit and post them, which I will do this weekend. The net is that like the other time I installed Feedland in my home lab, I could not get browser access via HTTP to work. The web browser just simply refuses to load the site and displays an SSL error message. The good news is that I did get access to work via my Nginx and duckdns.org configuration in the same manner as I have done with other apps hosted in my home network.
I have moved some equipment around in the home office so that I can use the adjustable/standing desk while standing. This is the first time I have experienced moving my Macbook between external monitor configurations and I am impressed by the fact that it remembers the different configurations.
I really hate how streamers create horrible user experience for the sake of commercials. I am watching the Olympics on Peacock and using the “Top Events” multi-view mode to watch a hockey game, when the period ends I want to check in on the other events but cannot switch because Peacock disables navigation during commercials. How do they think people use multi-view?
Tigers and Justin Verlander agree to one-year contract for 2026 season
Really glad to see Verlander return to where it all started for him. Who knows how much can contribute but he does have experience to pass along.
A benefit of retirement is that I am able to watch the 2026 Winter Olympic events live during the afternoon and because of that I am seeing much more of the Olympics than ever before. In addition to having the time to watch, technology, particularly Peacock, is helping by putting all of the U.S. coverage in one spot. To be honest, I am not constantly watching but I have it on in the background as I sit here at my desk.
Today for the first time in a month I took a walk outdoors in sweats rather than my full winter gear. Temps are already in the high 30s and could reach the high 40s. And, today pitchers and catchers report to MLB spring training.
Down The Drummer Rabbit Hole
Earlier today I noted that Dave had posted about the issue I encountered on January 30 preventing me from publishing my Daynotes using the “default” blogging tool in Drummer referred to as “Old School,” which is descriptive of a day based blogging format.
I started the process of making and testing the changes that Dave suggested, finding them to be more involved, but the net result is that the Daynotes site is now served by HTTPS rather than only being accessible via HTTP.
When the ability to build the blog from within Drummer broke I did my own work around by creating a new OPML file that is served by my instance of PagePark, which is Dave’s web server app that knows how to render OPML files for web browsers. For now I am going to continue using my “new” Daynotes outline file while using the Drummer blog for longer articles. My simple forwarding url for daynotes (daynotes.frankm.info) will go to single outline page while I have created a new forwarder for the Old School site (oldschool.frankm.info).
As I have been blogging more about tablets, particularly e-ink tablets, I added a Tablets category here and today I spent some time adding that category to related posts I have written during the last year.
Appears that Dave is now aware of the problem with the Old School blogs. He says the real issue is that the blogs are now served by https and the template they are based on does not support that. He provides instruction for how to fix, which starts by downloading a minimal template, editing it, putting the template html file on a public server, and then adding a header variable to the blog outline. Unfortunately at the moment Github is down so I cannot work on the fix, which for me will involve hosting the template file on my shared files host that is built from a Github repo. More details are in my Daynotes outline.
Analyze Oil Consumption Using NotebookLM
Of all the AI tools currently available I use Google’s NotebookLM the most. My best way to describe NotebookLM is that one can use it to apply Google’s Large Language Models toward a topic based on sources for information that one provides. For example, when we were researching health insurance plans last fall I created a notebook in NotebookLM and uploaded to it as sources PDFs provided by the insurance providers. I then used the chat in the notebook to ask questions about the different plans, which I then saved for future reference.
I have a 2013 GMC Terrain and a few years past I learned that this make and model SUV has a history of burning oil, so I have been diligently checking the oil level once a week and after long drives during trips. I log the date, odometer reading, and oil level in a note in Google Keep using my phone and I wish Gemini were integrated in Keep so that I could directly ask questions like how many miles have been driven and when is the last time I added oil.
Google has not added Keep as a source for NotebookLM, which I think would be logical, but what I can do is send (export) a note in Keep to a Google Doc which I can then add as source for NotebookLM. Today I exported each of my three log notes to a Google Doc, created a notebook in NotebookLM and add these docs as sources. NotebookLM correctly identified the sources as a maintenance log for a vehicle but it didn’t know which type of vehicle nor did it know the year the entries started because I only recorded the month and date. I created a README note that I added as a source in which I specified that the first date was in 2024 and that the logs are for a 2013 GMC Terrain. I also added some information about the dipstick markings.
With the oil logs in NotebookLM I am able to ask simple questions like, how many miles were driven between the last oil check and the prior check? I asked NotebookLM how many miles were driven in 2025, and it then offered to make an infographic that broke down the 2025 mileage by month. NotebookLM also created a detailed vehicle usage and oil consumption analysis report.
NotebookLM provides me with a simple and powerful way to analyze this data, but the fact that Google Keep cannot be a source means that as I update the current log note I will have to re-export it to the Google Doc and refresh that doc source in NotebookLM. I hope that in the future Google will add Keep as a source to either Gemini on my phone or to NotebookLM.
Finished reading: Behold the Spirit by Alan Watts 📚
Expectations For The Pixel 10
I got the Pixel 10 for Christmas. We ordered it from Best Buy during their “Black Friday” sale at a cost of $549, when launched this phone cost $799. With an $80 trade-in of a Pixel 6a the total cost to me is $469, which is a very good price for a “flagship” phone.
Before I received the Pixel 10 I wrote this expecting it to be a blog post but I never published it, so now I am going back and I am going to put in quote format my original expectation and then my current point of view after one month of use.
Yesterday, in response to a post by Manton about an approach to link rot that he is trying out, I speculated that utilizing the Internet Archive might be a better approach. Turns out, there is a Wordpress plugin that does this very thing.
My first experiment with a mechanical keyboard was brief, I returned the Keychron C3 Pro because its height messed up the ergonomics of my desk and my wrists hurt when using it. I normally use the Logitech MX Keys Keyboard that has a very low profile. For some reason I didn’t think about the additional height of Keychron keyboard, to use it properly I would need to elevate my chair high enough that my feet wouldn’t touch the ground.
Blogbook is an app in development that one can use to make PDF and EPUB files by exporting from WordPress, Micro.blog, or Ghost blog to a single Markdown file. Filter by categories, authors, tags, and dates–then open in Marked 3 for PDF, EPUB, and more. I can create a physical book from an export of my micro.blog to Day One, but that takes all posts in a date range while what I really wanted to do was only publish in book format essays.
Avoiding Dead Links
I currently subscribe to the basic micro.blog plan and I have been monitoring the features that Manton adds to premium for any that I might find useful. Today Manton posted a video demonstrating a feature he is experimenting with that could be added to premium that would help prevent dead links.
A problem that exists with adding links to other sites in my blog posts is that over time the pages I link to disappear. Chances are good that if you click a link on a post that I wrote five years ago the link no longer exists, which is a bit of a pain.
The new feature that Manton is experimenting with enhances the Bookmarks feature of micro.blog that creates an archive copy of a web page. Booksmarks is similar to Archivebox. The enhancement associates the original source URL of a archived page that one may use in a blog post and provides a “single click” way to convert the URL to the archived copy of the page on micro.blog and then you can update the post.
Of course, the problem with the current approach is that one has to know that the source URL is no longer available and make the change. For this to really be useful there needs to be some form of automation, which I can imagine could increase the costs of running micro.blog.
Perhaps a compromise is providing micro.blog users a method to initiate a scan of archived posts for dead links, check for the ones that are dead, see if there is an archived version and offer to convert the link. Such ad hoc scans might result in lower compute costs.
Another thing that would useful, if this becomes a real feature, is provide a way for the micro.blog user to an archived copy of the linked to page at the time of writing the post. I don’t know if Manton would want to automatically create an archive of all destination pages of blog posts, but that would be useful for people like me who often write posts using an editor other than micro.blog’s.
A final thought…. I wonder whether this could be integrated with the Internet archive is some way? It seems the purpose of the Internet archive is to do this type of web page archival and their storage may be a way minimum the costs to micro.blog.