A ton of digital housekeeping on a Saturday

To prepare for the annual publication of The Lindahl Letter book I went ahead and loaded up the posts from week 50 to 77. That involved doing the typesetting for roughly 25,000 words that spanned about 100 pages. I’m considering either bundling two years of posts together or just publishing one year at a time. 

I wrote the social media copy for sharing to Twitter and LinkedIn this week:

You can check out week 74 of my weekly technology newsletter The Lindahl Letter titled “ML content automation; Am I the prompt?” over on #Substack via the link below

—> Tags: #MachineLearning #ML #ArtificialIntelligence #AI #MLOps #AIOps

https://nelslindahl.substack.com/p/ml-content-automation#details

It got shared over to Twitter here:

The LinkedIn post is here:

https://www.linkedin.com/feed/update/urn:li:activity:6946447208812613632/

I had posted weekly into Facebook as well, but that yielded very little traffic back to the actual Substack site. After a few weeks of that effort I gave up and just stopped that part of my weekly social media sharing routine. I’m not entirely sure that sharing on Twitter and LinkedIn results in any subscribers, but it does create a small amount of traffic each week. 

At some point today, I’m going to work on writing the “Substack Week 78: Why is diffusion so popular?” post to create a draft. That is all queued up as a task for the day.

Oh that Wednesday arrived

Posts can now resume with the great server migration completed. My thoughts over the last two weeks have been confused on things, but not on writing or chronicling my efforts along the search for the perfect possible path forward. During the course of the brief online freeze I did use Twitter a bit and that was an interesting journey. It appears that Twitter has been purchased and will become a privately held company. It made me wonder about the total value of stock for a company. Generally speaking at the given market price of a stock everybody could not sell their stock. At the point where everybody that held a stock was trying to sell it the value of that stock would plummet exponentially. Most companies in the marketplace could not support at their current value a total replacement of investors. I actually find it interesting to look at the amount of a stock volume that trades hands every day for a given company. For the most part, you will see a small volume of stock up for trade each day will be far less than 20% of total outstanding shares. 

You can go look at the biggest average daily trading volumes (ADTV) for companies. Yesterday on April 26, 2022 for example the largest daily trading volume was for Twitter at around 115 million shares. The total number of outstanding shares for the company is around 763 million. That amounts to around 15% of the outstanding shares for the company being exchanged yesterday. In general, that is a large amount of activity between buyers and sellers for a single company. My thoughts then drifted into thinking about those metrics for a bit before circling back to the activity that is supposed to be at hand which is writing this morning. I had wanted to write 1,000 words in this very word processing document before sending this post off for publication. That is where my head was at this morning. I figured producing a good block of prose would help start my day off on the right path. 

I had brewed a cup of the Ethiopia Nespresso pod coffee instead of two shots of espresso today. Sometimes you have to mix it up a little bit. I’ll probably order those pods again at some point. I have been enjoying them as an alternative to two shots of espresso. My afternoons of course should be green tea and that is how the plan is supposed to work. Coffee or espresso in the morning and green tea in the afternoon. Sometimes the green team simply gets replaced with more espresso. That happens from time to time depending on what is going on during the day. For the most part being focused on the work at hand becomes the priority and settling into a state of deep focus on the hard challenges is where things should drift to after the start of the day. The first few minutes of the day are generally cloudy and it takes just a little bit to go from that fog of daybreak to a truly productive state. 

Even with that little bit of prose about coffee this post is only about half way to the targeted work count for this stream of consciousness based effort. Part of the desired process is to just let my mind wonder in whatever direction happens to be pulling my attention. At the moment, my attention is nearly wholesale focused on a task list and working toward the resolution of the things on that list. Everything else might as well be far off into the distance like shadows of things you could approach, but it would require a lot of effort to get to that point. Something will have to give on that front and hopefully it will be a burst of productivity aimed at reducing the task list to a manageable amount of things. That should open the door to some pursuits outside the process of revisiting a list for the next item. Sometimes you have to move beyond the get next command and ask yourself about the things occurring in the wild space outside that list. 

Just for content if you were trying to produce a solid million words per day instead of working on a 1,000 word start of the day posting production target. You would have to work a little harder and probably invest a bit more time into the production target of 2,740 words per day. Given the reality of sustaining that level of productivity you probably will want to work toward a production level of posting target of 3,000 words per day. Sitting down and writing 3,000 words in a single sitting or flourish of creativity is possible. Doing it for an entire year is a different matter altogether. That type of sustained writing productivity is geared more toward something that a professional writing would achieve. Producing a million words in one big giant pile of prose for any reason is an awfully large amount of content. Very few people sit down and try to engage in that level of production. You could very well say that it would be a big year. A categorically larger than what you would expect to the point of being you considering it a “big” writing productivity year. That point is probably not worth any more inquiry. It has been examined and can probably stand on its own merits at this point. 

My closing thoughts on this writing session are all over the place at the moment. I’m thinking about trying to take on a big year again. At the same time, I’m remembering that that is a very big investment of writing time and energy. Sitting down and writing 3,000 words without an outline or any guide posts of what should be covered is a really intense commitment into the world of stream of consciousness prose. At that point the final outcome of what is created will be a consequence of the process, not the achievement of something on a task list.

Maybe I should live Tweet today

Today could very well be a live tweeting extravaganza. Two shots of espresso were brewed using a Nespresso machine. 400 calories of chocolate flavor Huel were consumed. Now that I’m all fueled up, I have a couple hours to write.

Today I got left behind during spring break week. Apparently, all the fun is happening elsewhere and it was time for me to be left back at the basecamp for writing and reading. Today is @NVIDIA CEO Jensen Huang’s keynote. It was not a bad day to get left behind. 

I finished the upcoming “Language models revisited” post and should be focused on writing a Substack post about “Ethics in machine learning” with my extra time today. I’m looking at other content creation options, but will probably swing back to generating 1,000 words on ML ethics.

A little music was required to begin the writing process. I’m listening to my Warren Zevon station on Pandora. It was apparently created back on May 5, 2013. You can find it here: ? Warren Zevon Radio on @PandoraMusic https://pandora.app.link/hMZALDTGBob

Things are about to get going with Jensen from @NVIDIA. I have the keynote ready to go in a browser tab “GTC 2022 Keynote [S42295]” is waiting to play. Don’t panic with FOMO. It has an indicator of what will be covered, “Primary Topic: AI Strategy for Business Leaders.”

The GTC event has so many sessions it would take a lot of time and effort to listen to all of them. I’m thinking beyond the keynote I might attend, “A Vision of the Metaverse: How We will Build Connected Virtual Worlds [S42114].”

Ok the #GTC22 player failed with the error message “Something went wrong” from the actual @NVIDIA event player and I ended up going over to @YouTube to watch the actual keynote live stream with 6,713 other people watching the premiere. https://www.youtube.com/watch?v=39ubNuxnrK8

This keynote from Jensen is super choreographed and heavily embedded with graphics. It’s almost overproduced at this point. I’m getting a bit of infomercial flavor compared to the technical content I wanted to consume. However, #GTC22 is holding my attention so far today.

This #GTC22 keynote is really full of machine learning buzzwords. Anybody playing bingo to this keynote probably has already had a winner. They really had Jensen reading a ton of rather dense lists during the recording process.

I was really hoping Jensen would announce they built an actual Jurassic park during #GTC22. Would they have to pay Michael Crichton a royalty?

I’m curious about how the GPU confidential computing will work in practice. The #GTC22 keynote has been interesting so far.

Whoa… That is a fast computer from @NVIDIA. The #GTC22 Eos will be 275 petaFLOPS or 1.4x faster than the fastest science computer in the United States…

NVIDIA Eos screenshot

Like Jensen I’m very curious what the next million-x will look like as well. #GTC22 should be interesting this year. I’m gearing up for the “A Vision of the Metaverse: How We will Build Connected Virtual Worlds [S42114]” panel that is starting soon. The videos should be here: https://www.youtube.com/playlist?list=PLZHnYvH1qtOYXzWxVdIU1ZDpbLvxbZdyQ

Writing in Tweets

Maybe my weblog post today should be drafted in sections that are the length of a maximum Tweet at 280 characters. That will shrink my paragraphs by a lot for this post. I’m considering just composing a bunch of blocks and then posting them on Twitter as I write some prose today.

My writing plan to produce a Substack post each week has been working. Week 42 about time crystals has now been written and will go live in 6 days. In 10 weeks, I will have hit my goal to publish a weekly machine learning based post for an entire year. https://nelslindahl.substack.com/

Earlier this week I moved my office desk from my downstairs office. I did spend a majority of my time in that office sitting in my blue Scandinavian Design’s Wau desk chair. My desk and that chair are now upstairs for a few weeks. A majority of my time will now be spent upstairs.

My internet modem is now upstairs. I was worried that the gigabit speed would be diminished by the house wiring vs. a direct run to my office. Comparatively the speed loss was 20% from the top end. That is acceptable for now. My plan in a few weeks is to move back to my office.

Editing the idea I want to communicate from each paragraph down to 280 characters really slows down my writing output. Maybe it will help me be more word economic and succinct go forward. My writing is not really overly wordy to begin with so editing down requires real effort.

A little bit of daily writing

For the last two days of posting on this weblog the content has been sent over to Twitter as a thread (aka Tweet storm). It still feels like the handling of threads by the Twitter application/website is somewhat interesting in practice. You can end up seeing or looking at the middle or end of a thread and the end user sort of has to decide to view the thread in totality. It does show some indicators that you should click on the very happy “Show this thread” indicator at the bottom of any Tweet, but it seems rather shoehorned to fit or an artifact of after the fact design. Naturally, I’m going to keep going down this road for at least the next few posts to really get a feel for how the feature really works out in the wild. Neither of the Tweet threads sent any traffic back to the weblog, but the WordPress reader did send over a few readers who were probably interested in my thoughts on automation. That is a topic that is definitely worthy of consideration. 

My efforts to return to the habit of daily writing at the start of the day are working out well enough. It all starts with waking up early enough to get things done without anybody else in the house moving around. That is probably why the time at night and the time at the start of the day are the best use cases for producing epic tomes of lofty weight and meaning. It certainly takes a bit of energy to engage in the active practice of reflection along the way. I have a path that involves dedicating a few steps forward every day to achieve, but the actions and tasks between those steps are the stuff that defines my productivity. Both a day and a stone stand the test of time. I’m probably not going to do any stone working so my aim has to be to make the most of the time comprising my day. That is what sitting down and wrestling with the blank page is all about. It is about creating prose from the top to the bottom of the page and realizing the journey of doing that unfolds a world of through experiments that otherwise would have gone without the spark to create them. That is really the natural output of conducting this type of writing output on a daily basis. The goal of the process is to create the opportunity for something interesting to spring into existence.

Experiments: 1) For this post I set a featured image to see what it would do to the Twitter thread. 2) I also included a link to Twitter in the second sentence to see how it handles external links in the thread process.

Thinking about that automation

My first weblog to tweet storm automation went well enough. It did not translate to any new weblog traffic of any kind, but it was sort of fun to see the post get broken up via automation and sent out a series of linked tweets in thread form. For the most part this went like most tweets it was nothing more than an utterance in the grand public commons that went without notice or observation. It happened and that is about all that needs to be said about it at this time. I’m going to keep using this method to post content and see how it goes for the next few days. It does require a little bit more work at the time of posting, but it is sort of fun to see it populate the tweets at the time of publication. It is interesting how much automation is being built out to cross post and to share content across a multitude of platforms. It is probably an artifact of the thinning level of attention each application or forum has within the grand context of the public mind. In the aggregate the attention of the people has been diluted to a point where getting a message out to the vast majority of people is actually really difficult to achieve in practice. To that end the use of automation to reshare and send content flying from platform to platform is probably going to only increase the level of dilution as the sharing increases in velocity. 

One of the things that I truly fear is that models like GPT-3 and others will be used to create so much content within the stream that everything else will be drowned out. It is entirely possible that the classical method of news reporting could be overwhelmed by bad actors using a firehouse of near real content created by GPT-3 to simply make finding the signal within the noise an impossible act. That is a fear shared by other NLP and machine learning researchers who know that the use of language models to turn out endless streams of content could overwhelm our current channels of communication. That thought should probably end up in a Substack newsletter instead of a weblog post, but I’m still struggling with the mechanics of doing both of those things at the same time. Sometimes my start of the day writing sessions are going to be more geared toward that type of content instead of pure stream of consciousness. It is entirely possible that my thoughts of the moment will drift into the production of more technical and content driven areas. When you go where your whims will take you it could be toward more academic writing or it could be toward the passions of the moment. That is the nature of that type of writing effort. 

I am still sitting down and trying to fill up one blank page of content in the morning at the very start of the day. It is pretty easy to just open up Google Docs and start writing from the top of the document and watch the content fill up the screen. I did end up getting a Kensington Duo Gel Keyboard Wrist Rest to go with my 60% size K65 Corsair mechanical keyboard. Typing for a long stretch without the wrist pad was not very comfortable and just adding the gel wrist rest was enough to make things feel much more comfortable when typing. That probably has something to do with ergonomics and the mechanics of proper tying, but my explanation is grounded in the practice of knowing one way was uncomfortable and the other allows me to type for longer stretches of time. This post will go online here shortly, but it needs to be read and revised before that happens. The word count on this batch of prose is well over 500 so it was a good start to my planned daily writing activities.

Setting up a bit of automation

Well I have moved back into the pattern of writing a daily weblog post. Each morning I’m sitting down and producing a few words in a near stream of consciousness fashion. This is in addition to my efforts on Substack each week. Along the way, it seemed like a good idea to update my “Publicize posts” setting to create links from Twitter to automatically share new weblog posts by title and a link when they are published. Oddly enough they have a new setting since the last time that I used that functionality. Apparently, I can set up each post to compile and publish as a thread on Twitter, “Twitter Thread: Share the content of this post as a Twitter thread.” I’m kind of curious what it does with multiple photos in the same post. I’m wondering if it just posts text or if links, images, or embedded videos will be compiled over. Some of the documentation says that it will bring over all the content. That is something that I will have to test out by embedding images and maybe some links or something to this post.

I ate the pretzel in this photo. It was delicious. I’m curious how it will show up in the Tweet thread. Right after this line of text is a photo of a giant pretzel with a cup of cheese and a cup of red sauce.

The pretzel was delivered with this extra large New York style cheese pizza. Right after this line of text is a slightly off to the right picture of that pizza.

Here is the output of what happened after publication. It made a thread of Tweets, but the images did not carry over it sort of just compressed the content into a series of linked Tweets.

Thinking about social networks

My social media presence includes Twitter, LinkedIn, and Substack. The other day I did login to Facebook, but that was more a one off than anything social media related. For the most part LinkedIn is a type of currated living resume vs. any type of real social network. Some people do seem to use it in a more social way, but that is rare. Maybe it can be in some ways viewed as a replacement to the business pages of the phonebook. 

A note on phonebooks: In the olden days we used to get physically printed lists of phone numbers delivered on paper to the door of homes. This was not voluntary and it included a special section or sometimes an entirely separate book for business phone numbers. If this post is being far enough into the future that phone numbers are inherently confusing as a concept, then you should probably just skip this post and read something else. Anyway, LinkedIn is a very odd look into the way people want to be seen online in a busines context. 

Twitter on the other hand is really just an open town square of some loud, but mostly inconsequential utterances. The vast majority of tweets are seen by very few people and are of little consequence. You do get a feel for what the crowd is thinking or doing at any given moment, but it is at times highly chaotic. I keep thinking that a curated section will develop where the most powerful of the blue check marks will assemble into an echo chamber of thought experiments. That has not really happened in any sustained or meaningful way.