Earlier this week I was thinking about what exactly happens at the end of the blog. Most of the time in the lifecycle of a weblog or blog the end happens from abandonment. Probably the vast majority of blog type writing projects have been just abandoned. At some point, the writer just stops producing that type of prose and moves along to something new. A few of them were powered by writers that sustained them for years or perhaps decades. Those platforms of prose generation stood the test of online time. Generally, at the point of abandonment most of the self hosted blog experiments eventually vanish, expire, or are terminated. Sometimes they were built on a platform that just sustains and lingers. Those free platforms sometimes can last a very long time in the online world.
In my case, from this point I know that the servers are paid out 5 years from now and assuming the platform properly updates itself the blog could survive during that time frame. Certainly the prose won’t really improve during that time. It will just survive online. My plans at the moment are to keep adding to the content. I write for the blog without consideration for an audience. The content is created really for my own purposes of writing. Throughout the last 20 years the blog content just mostly sits, lingers, and remains unmoving and uncompelling. It’s writing without a discrete future purpose. The prose was formed within the process of writing.
Considering some writing schedule updates:
Saturday – daily blogging, early morning hours spent on The Lindahl Letter development
Sunday – daily blogging, early morning hours spent on The Lindahl Letter podcast recording
Monday – daily blogging, nels.ai development
Tuesday – daily blogging, nels.ai recording
Wednesday – daily blogging, nels.ai publishes at 5 pm
Thursday – daily blogging, big coding adventures
Friday – daily blogging, The Lindahl Letter goes out at 5 pm
I have the outline of a book that probably needs to be written sometime soon. I could devote my Saturday and Sunday early morning time to working on the chapters of that book as blocks of content creation. All of that content is listed in the backlog and will eventually get built, but maybe the time to produce a certain section of that backlog is now instead of leader. It’s always the reframe of action that the time is now. Finding and sustaining the now is probably the harder part of that equation.
Today started off in a rather normal sort of way. Two shots of espresso were made and were delightful. Sunrise happened outside the view of my window. My Saturday morning routine of watching a bit of the WAN show happened without interruption. I took a few moments to review my top 5 things from yesterday and it is somewhat satisfying to review and consider the flow of things from day to day. Being a reflective builder is an important part of the process. My argument represented as a hypothesis would be that on any given day we can accomplish 5 blocks of time building good things. To me that is a reasonable way to look at building and creating. Some people for sure are able to work in a different way creating more or less blocks of production. Generally I’m looking at reasonably hard things that are broken into achievable blocks of things that can be done. I cannot code a whole application in a single block of time. That task could be broken into a reasonable set of blocks and I could certainly work on completing that effort.
Right now I’m working to finish up block 142 of the Lindahl Letter Substack publication. I’m seriously considering closing the newsletter at 150 weeks of writing effort. I might let it go till 156 weeks which would be a complete 3 years of content generation. I had considered switching to a pay model and delivering more in depth independent research each week. Each week right now I provide a brief research note on the topic I’m interested in researching. It’s really a sharing of what I’m interested in and that is the sole and direct focus of the writing enterprise on that one. I have already moved to sharing the same content on my weblog each week at the same time. That got me thinking about where people consume content these days.
Within academic spaces content has always been harder to access than it should have been with paywalls, high prices, and subscriptions. Journals are great for keeping and storing ideas shared between academics who subscribe and read the journal. It’s a community of interest and it works generally for that academic community. People outside that circle wanting access might need to go to a library or decide if they want to pay for the journal. It’s a limiting circle of content management. Publishing a series of research notes is probably essentially ephemeral in nature. While in the abstract the internet never forgets we have reached the point where it’s really large and probably not backed up. That ephemeral nature will mean that the weekly posts will probably at some point vanish. I had considered that reality from the start of the endeavor and at the end of each year I pooled that year’s Substack content into a book. Right now two of those ponderous tomes of thought sit next to me on the shelf.
Those efforts will probably stay in publication longer than anything stored on the internet at large. I keep my web hosting paid for 5 years out so in theory that is the longest horizon of serving up that content on the open internet. I’m digging into some deeper topics today and that is interesting for a Saturday morning.
Thank you for tuning in to this audio only podcast presentation. This is week 138 of The Lindahl Letter publication. A new edition arrives every Friday. This week the topic under consideration for The Lindahl Letter is, “Prediction markets & Time-series analysis.”
We have been going down the door of digging into considering elections for a few weeks now. You knew this topic was going to show up. People love prediction markets. They are really a pooled reflection of sentiment about the likelihood of something occuring. Right now the scuttlebut of the internet is about LK-99, a potential, maybe debunked, maybe possible room temperature superconductor that people are predicting whether or not it will be replicated before 2025 [1]. You can read the 22 page preprint about LK-99 on ArXiv [2]. My favorite article about why this would be a big deal if it lands was from Dylan Matthews over at Vox [3]. Being able to advance the transmission power of electrical lines alone would make this a breakthrough.
That brief example being set aside, now people can really dial into the betting markets for elections where right now are not getting nearly the same level of attention as LK-99 which is probably accurate in terms of general scale of possible impact. You can pretty quickly get to all posts that the team over at 538 have tagged for “betting markets” and that is an interesting thing to scroll through [4]. Beyond that look you could start to dig into an article from The New York Times talking about forecasting what will happen to prediction markets in the future [5].
You know it was only a matter of time before we moved from popular culture coverage to the depths of Google Scholar [6].
Yeah, you could tell by the title that a little bit of content related to time-series analysis was coming your way. The papers being tracked within Google Scholar related election time series analysis were not highly cited and to my extreme disappointment are not openly shared as PDF documents [7]. For those of you who are regular readers you know that I try really hard to only share links to open access documents and resources that anybody can consume along their lifelong learning journey. Sharing links to paywalls and articles inside a gated academic community is not really productive for general learning.
If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. New editions arrive every Friday. Thank you and enjoy the week ahead.
Focusing inward on delivering requires a certain balance. My balance has been off recently. I got knocked off my feet and it impacted my ability to produce blocks of content for about a week. That type of thing does not normally happen to me. It was a new set of emotions and things to consider. Getting knocked down hard enough to pause for a moment and need to look around before moving again was a very new sensation. I’m not sure it was something that I was looking for or even prepared to experience. Really the only thing that put me back on the right track to success and deeper inward consideration (restored balance) was the passage of some time. It just took a little bit of time for me to internalize and move on to a new set of expectations.
Each new day brings forward a set of time for creating blocks of content. My thoughts right now are around the consideration of making and maintaining 5 slots of building each day. To that end I have been sitting down on the whiteboard and writing down 5 good things to work on each day and trying to make sure they are attainable blocks to complete. At this time, I don’t want to put multi-slot blocks or all day blocks on the board for action and review. This is not the time for that type of stretching and personal growth by taking on highly complex activities. Right now is the time to make things clear, work on the clear things, and be stronger with that resolution every single day go forward.
Maybe getting back to the absolute standard of sitting down at the very start of the day after drinking two shots of espresso and writing for a few minutes is the key to reframe my day. It is something that has been missing. It was missed. Perhaps it was missed more than I even realized at the time. I’ll admit to sitting down and watching about 4-5 seasons of the Showtime series Billions instead of actively writing and building. Alternatively, I could have been listing some graded sports cards on eBay and working to sell a few of them each day. Let’s zoom out for a second from those thoughts and consider what the next 30 days will uphold as a standard.
One block of the daily 5 is going to be related to committing code on GitHub. I’m going to really focus my time and energy on making solid contributions to published code. Taking on that effort will help me be focused and committed to something that will become more and more necessary. Building code has changed a bit with the advent of LLMs, but the general thought exercise and logic remain pretty much the same. You might be able to take a wild run at something that was not attainable before and prompt your way to something magical. Generally you are going to go where logic can take you within the confines of the coding world as the framework is a lot more logical than it is purely chaotic in nature.
Outside of those efforts generally as a part of my daily routine I’m producing a daily vlog via YouTube Shorts and striving to output a daily reflection functional journal blog post. I’m going to try to take some inline functional journal notes throughout the day as well. That is going to structurally end up with a sort of blog post being written at the start of the day and then a bunch of more inline bullets being created. Posting is still going to happen at the end of the day or potentially a day delayed.
Delivering independent research is more important now than ever. I have spent some time thinking about the models of how that research is delivered and what value it has generally.
Block 142 is pretty much ready to go. I’ll be able to record it tomorrow morning and stay on track to have a 4 block recorded backlog of content ready to go for my Substack.
During the course of reviewing blocks 143 to 145 I considered if those are even the right topics to spend time working. They are probably fine elements of things to research. It’s not about producing timely content, but instead it is about making meaningful blocks of content that are not time sensitive. That of course is always a harder thing to accomplish while producing independent research.
Thank you for tuning in to this audio only podcast presentation. This is week 137 of The Lindahl Letter publication. A new edition arrives every Friday. This week the topic under consideration for The Lindahl Letter is, “Tracking political registrations.”
Trying to figure out how many republicans, democrats, and independents are registered in each state is actually really hard. It’s not a trivial task. Even with all our modern technology and the extreme power of the internet providing outsized connectedness between things and making content accessible to searches. Even GPT-4 from OpenAI with some decent plugins turned on will struggle to complete this task.Your best searches to get a full list by state are probably going to land you into the world of projections and surveys. One that will show up very quickly are some results from the Pew Research which contacted people (300 to 4,000 of them) from each state to find out more data about political affiliation [1]. They evaluated responses into three buckets with no lean, lean republication, or lean democrat. That allowed the results to evaluate based on sampling to get a feel for general political intentions. However, that type of intention based evaluation does not give you a sense of the number of voters within each state.
It opened the door to me considering if political registration is even a good indicator of election outcomes. Sports tournaments rarely play out based on the seeding. That is the element of it that makes it exciting and puts the sport into the tournament. To that end back during week 134 I shared the chalk model to help explore a hypothesis related to registration being predictive. At the moment, I’m more interested to see how proxy models for predicting sporting events are working. Getting actual data to track changes in political registrations is an interesting process. ChatGPT, Bard, and Bing Chat are capable of providing some numbers if you prompt them properly. The OpenAI model GPT-3.5 has some older data from September 2021 and will tell you registered voters by state [2]. I started with a basic prompt, “make a table of voter registration by state.” I had to add a few encouraging prompts at some points, but overall the models all 3 spit out results [3]. The Bing Chat model really tried to direct you back to the United States Census Bureau website [4].
This is an area where setting up some type of model with a bit of agency to go out to the relevant secretary of states websites for the 30 states that provide some data might be a way to go to build a decent dataset. That would probably be the only way to really track the official data coming out by state to show the changes in registration over time. Charting that change data might be interesting as a directional view of how voters view themselves in terms of voter registration in a longitudinal way. People who participate in Kaggle have run into challenges where election result prediction is actually a competition [5]. It’s interesting and thinking about what features are most impactful during election prediction is a big part of that competition. Other teams are using linear regression and classification models to help predict election winners as well [6]. I was reading a working paper from Ebanks, Katz, and King published in May 2023 that shared an in depth discussion about picking the right models and the problems of picking the wrong ones [7][8].
To close things out here I did end up reading this Center for Politics article from 2018 that was interesting as a look back at where things were [9]. Circling back to the main question this week, I spent some time working within the OpenAI ChatGPT with plugins trying to get GPT-4 to search out and voter registration by state. I have been wondering why with a little bit of agency one of these models could not do that type of searching. Right now the models are not set up with a framework that could complete this type of tasking.
If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. New editions arrive every Friday. Thank you and enjoy the week ahead.
Thank you for tuning in to this audio only podcast presentation. This is week 136 of The Lindahl Letter publication. A new edition arrives every Friday. This week the topic under consideration for The Lindahl Letter is, “Econometric election models.”
It has been a few weeks here since we started by digging into a good Google Scholar search and you know this topic would be just the thing to help open that door [1]. My searches for academic articles are always about finding accessible literature that sits outside paywalls that is intended to be read and shared beyond strictly academic use. Sometimes that is easier than others when the topics lend themselves to active use cases instead of purely theoretical research. Most of the time these searches to find out what is happening at the edge of what is possible involve applied research. Yes, that type of reasoning would place me squarely in the pracademic camp of intellectual inquiry.
That brief chautauqua aside, my curiosity here is how do we build out econometric election models or other model inputs to feed into large language model chat systems as prompt engineering for the purposes of training them to help either predict elections or interpret and execute the models. This could be a method for introducing extensibility or at least the application of targeted model effect to seed a potential future methodology within the prompt engineering space. As reasoning engines go it’s possible that an econometric frame could be an interesting proxy model within generative AI prompting. It’s a space worth understanding a little bit more for sure as we approach the 2024 presidential election cycle.
I’m working on that type of effort here as we dig into econometric election models. My hypothesis here is that you can write out what you want to explain in a longer form as a potential input prompt to train a large language model. Maybe a more direct way of saying that is we are building a constitution for the model based on models and potentially proxy models then working toward extensibility and agency from introducing those models together. For me that is a very interesting space to begin to open up and kick the tires on in the next 6 months.
Here are 6 papers from that Google Scholar search that I thought were interesting:
Beyond those papers, I read some slides from Hal Varian on “Machine Learning and Econometrics” from January of 2014 [2]. The focus of the slide was applied to modeling human choices. Some time was spent on trying to understand the premise that the field of machine learning could benefit from econometrics. To be fair since that 2014 set of slides you don’t hear people in the machine learning space mention econometrics that often. Most people talk about Bayesian related arguments.
On a totally separate note for this week I was really into running some of the Meta AI Llama models on my desktop locally [3]. You could go out and read about the new Code Llama which is an interesting model trained and focused on coding [4]. A ton of researchers got together and wrote a paper about this new model called, “Code Llama: Open Foundation Models for Code” [5]. That 47 page missive was shared back on August 24, 2023, and people have already started to build alternative models. It’s an interesting world in the wild wild west of generative AI these days. I really did install LM Studio on my Windows workstation and run the 7 billion parameter version of Code Llama to kick the tires [6]. It’s amazing that a model like that can run locally and that you can interact with it locally using your own high end graphics card.
If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. New editions arrive every Friday. Thank you and enjoy the week ahead.
My ability to sit down and write a block of words has gone in and out of service in the last two weeks. That is understandable, but it still surprised me a bit that my focus could be disrupted. I have been absorbing a ton of new generative AI related technology. So many companies are spinning up in that space and they do all sorts of interesting things. An interesting element of that is how much of it runs back to a common set of APIs where people are calling models for answers. The interesting part of this equation is that people are spinning up local hosting and sometimes even hosting in notebooks.
I watched the Pinecone company sponsored, “AI Transformation Summit 2023.” They loaded the 11 videos as playlist here on YouTube:
Yesterday, I spent some time setting up my OBS, webcam, and microphone to record some desktop driven demos. It has been a few years since that was a priority for my efforts. Low friction workflows are super important to help drive my ability to create content. I want to be able to record and upload without really having to do any editing.
I watched this video and learned how to set up a special filter to change my webcam video from a rectangle to a circle in OBS. Strangely enough, this was an oddly rewarding filter to create. It was not a hard thing to do, but it was a little bit rewarding. For a few moments I did consider using the chat bubble for my filter. That consideration quickly faded away
Today got off to a slower start than intended. My alarm went off and I got out of bed and was ready to jump into working, but it took about 30 minutes to boot up and become productive. Even with two shots of espresso and some Sunday with Ola playing. Part of the slowdown was due in part to a bunch of web surfing about how Eight Sleep products work. We have two large dogs and I have been hesitant about any technology they could destroy. Yesterday, I started a little tracking project to see what applications or software packages I’m using actively right now. During the course of the last 24 hours of big adventures I just kept a list. It turns out that I used 15 different things. You can check out the list for yourself below. The list is roughly in the order of how the applications were used throughout the last day.
Android OS – Alarm clock on my Pixel 7 Pro went off at 5:30 AM
Windows 11 OS – unlocked my main computer the Dark Base Pro 900
Microsoft Outlook – Checked my primary email inbox
Gmail – Checked my secondary email inbox
Pokemon – I had to clear out some Pokemon to get ready for Go Fest 2023
PowerDirector – Edited some video clips for the daily vlog
YouTube – Watched the All-In podcast and the WAN show
Google Docs – edited some documents and did some writing
Google Podcasts – listened to some podcast audio during a walk
Pandora – listened to some music during a walk
LinkedIn – I’m checking this daily now
Twitter(X) – I’m still a daily active user of Twitter
YouTube TV – I watched a bit of the Bears vs. Bills game via Sunday Ticket
GitHub – I was messing around with some Colab notebooks
Google Colab – These launched out of GitHub
I probably used the Chrome operating system as well on my Pixelbook Go at some point during the day. Generally, I mess around with Colab notebooks on the Chromebook. The major GPU work on those notebooks is happening somewhere else so the Chromebook interface is plenty powerful to support the Colab notebook browsing experience. The list does not include any indexing to how many times I used anything. For the most part, I was just tracking the first use of something to help make the list. This was not intended to be a time in motion study of any kind. It was just an effort to help understand the tooling being used. The next step here would be to extend the sampling period to an entire week and see if anything else ends up being caught up in a longer sampling period.
Right now we are sitting at a recorded and processed backlog level of 3 weeks on the Substack. That 3 weeks of backlog is scheduled to run until Friday, September 15, 2023. Based on my normal routine for the weekend I should be recording week 139 right now during my Saturday morning. This morning The WAN Show was back (no comment) and I have the content ready to record for block 139. Instead of picking up the Marantz professional sound shield live vocal reflection filter from under my desk and starting to record on my Yeti X from Blue microphone I took a step back. Blocks 139, 140, 141, and 142 are actually written and ready to record, but they could be better. This morning I’m going to read all 4 of those blocks and give them a round of editing and expansion. That will be a part of the reworking content process that is important.
Recently, I have been way more likely to push the entire backlog back a few weeks and drop in new topics. That is in part a function of not wanting to put a new topic so far out in planning and review. Having a new topic that grabs my interest be shelved for over a year while the backlog is consumed in front of it just seems wrong. Right now that backlog goes out until block 226 which is really a problem for future Nels. Sometimes you have to be willing to mix things up and change up the backlog. It’s a healthy part of the agile process. I’m even going to create a new category for posts that is going to be about backlog management.
Throughout the next week I’ll be working with LangChain and building some Colab notebooks related to using and understanding all the use cases currently in that space. I’m going to spend time every day working on a single block of content within the backlog as well. That should end up yielding some good content building and a lot of focus on getting things done. For the most part the process of recording the audio, video, or other elements is distinctly separate from the writing process and is normally a part of my weekend routine.
Thank you for tuning in to this audio only podcast presentation. This is week 135 of The Lindahl Letter publication. A new edition arrives every Friday. This week the topic under consideration for The Lindahl Letter is, “Polling aggregation models.”
I read and really enjoyed the book by Nate Silver from 2012 about predictions. It’s still on my bookshelf. Strangely enough the cover has faded more than any other book on the shelf.
Silver, N. (2012). The signal and the noise: Why so many predictions fail-but some don’t. Penguin.
That book from Nate is sitting just a few books over from Armstrong’s principles of forecasting. A book that I have referenced a number of times before. It will probably be referenced more as we move ahead as well. It’s a resource that just keeps on giving. Math it’s funny like that.
Armstrong, J. S. (Ed.). (2001). Principles of forecasting: a handbook for researchers and practitioners (Vol. 30). Boston, MA: Kluwer Academic.
My podcast feed for years has included the 538 podcast where I listened to Nate and Galen talk about good and bad uses of polling [1]. Sadly, it does not currently feature Nate after the recent changes over at 538. They reported on and ranked a lot of polling within the 538 ecosystem of content. Model talk and the good or bad use of polling were staples in the weekly pod journey. I really thought at some point they would take all of that knowledge about reviewing, rating, and offering critiques of polling to do some actual polling. Instead they mostly offered polling aggregation which is what we are going to talk about today. On the website they did it really well and the infographics they built are very compelling.
Today setting up and running a polling organization is different from before. A single person could run a large amount of it thanks to the automation that now exists. An organization with funding to set up automation and run the polling using an IVR and some type of dialogue flow [2]. Seriously, you could build a bot setup that placed calls to people and completed a survey in a very conversational way. That still runs into the same problem that phone survey methods are going to face. I screen out all non-contact phone calls and I’m not the only person doing that. Cold calls are just not effective for business or polling in 2023 and the rise of phone assistants that can effectively block out noise are going to make the phone methodology even harder to effectively utilize.
It’s hard to make a hype based drum roll on the written page. You are going to have to imagine it for me to get ready for this next sentence. Now that you are imagining that drum roll… Get ready for a year of people talking about AI and the 2024 election. It probably won’t get crypto bad in terms of the hype trane showing up to nowhere, but it will get loud. I’m going to contribute to that dialogue, but hopefully in the softest possible way. Yeah, I’m walking right into that by reflecting on the outcome of my actions while simultaneously writing about them during this missive.
You can see an article from way back in November 2020 talking about how AI does show some potential to gauge voter sentiment [3]. That was before all of the generative AI and agent hype started. Things are changing rapidly in that space and I’m super curious about what can actually be accomplished in that space. I’m spending time every day learning about this and working on figuring out ways to implement this before the next major presidential election in 2024. An article from The Atlantic caught my attention as it talked about how nobody responds to polls anymore and started to dig into what AI could possibly do in that space, microtargeting, and Kennedy (1960) campaign references [4]. That was an interesting read for sure but you could veer over to VentureBeat to read about how AI fared against regular pollsters in the 2020 election [5]. That article offered a few names to watch out for and dig into a little more including KCore Analytics, expert.ai, and Polly.
We will see massive numbers of groups purporting to use AI in the next election cycle. Even The Brooking Institute has started to share some thoughts on how AI will transform the next presidential election [6]. Sure you could read something from Scientific American where people are predicting that AI could take over and undermine democracy [7]. Dire predictions abound and those will probably also accelerate as the AI hype train pulls up to election station during the 2024 election cycle [8][9]. Some of that new technology is even being deployed into nonprofits to help track voters at the polls [10].
If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. New editions arrive every Friday. Thank you and enjoy the week ahead.