It is currently October 1st in Calgary, Alberta, and I am Sarah Lacard, writing to you from 405 1133 8 Ave SW, on land stewarded by those humans indigenous to our lands from these perspectives, of particular and personal note, the Pikani of the Blackfoot Confederacy and Tribe(s), the tall people clan, and we give particular thanks to Sai'Kokoto for sharing their knowledge of the Blackfoot people with me and others from the Knox United Church congregation and attendees thereof.
Starting with convention, or rather, tradition and practices which I utilized when I was conducting such a practice previously, the song "Trouble Mind" started playing on my phone, which is presently between my arms while I type this on my computer workstation, facing south, within my studio apartment. I have decided to do some pomodoros of my own flavour and variety during the month of October, and beyond that, I am determining the proper, right, and doable while still comfortable shape of a new devotional. I will pray twice a day, in the morning, as I gain consciousness with my eyes closed and before I fully wake up, and this evening I will determine the best form of the shape of my evening prayers, as usually when I lay down to sleep, when I am ready to pass unconscious, I do so immediately without much thought, however it is likely that I will want to incorporate something more formal. One knee? Cross-legged? Kneeling fully? Sitting? Standing? Who knows—only time will tell, and certainly not a lot of time will be required for it to do so. I am also going to write or type for at least 25 minutes each morning, review what I wrote for approximately or around 5 minutes after that, do that once a day—every day—between 9am and noon, and then do something similar but more public between noon and maybe seven, and then maybe a third time between maybe nine and somewhere around midnight, and my plan is to maintain this for all of October. After editing, it looks like some of the details are vague, which is alright; they will be filled in as time passes.
Part of this practice is inspired or motivated (although both of those words attribute too much positivity and credit) by Visa's "one hundred things" that has been a task or assignment or challenge which has haunted me for many years since I first heard it and my subsequent and following inability to accomplish it at any point since then (mid-late 2010s), but I am going to take a final crack at it this month and through towards the end of the year and hopefully learn a thing or two from it in the process. Most generally, the reason why it has been such a difficulty is that it is hard to make something one hundred times, especially if you're trying to do that to an exact number or exact standard, or trying to make sure all of the one hundred things are consistently the same in some sort of way, but in this case, the loose structure of the devotionals I have prepared to do and set out to accomplish—albeit accomplish is certainly the wrong descriptor for such an undertaking—things are a lot more lax and loose and relaxed and not so strictly regimented or regulated or forced, and there is no sense of personal discipline required for me to accomplish it or them. I am allowing myself to adjust the shape of the task as they are completed, and even if I do complete two of these pomodoro posts for the month of October, then at most we'll be at around ninety-three of them, and as we continue on, the number one hundred should naturally arrive and pass without notice. Which is a good thing, in my opinion. And as a pre-school daycare teacher once told me (for some reason) it's not the ninety-ninth or one hundredth time that something goes wrong, but on the one hundred and first time where you will be punished for your misdeeds or careless actions. Who knows. I think it is best not to believe and hyperstition or woozle yourself or one's self into such an outcome.
Anyway.
With the start of October, it is also the start of the sprint for Canada's new AI Task Force—Canada now has an AI Minister, and they have put together a group or team of around twenty-eight people from my understanding in order to provide them feedback on the sort of things and the strategies that Canada should move towards, or something like that, so obviously, I am going to try and participate, and even though I am not a member of the task force, as far as I am aware they are also soliciting feedback from members of the public in this regard.
I have not looked at today's news or socials prior to writing this post, so it may be public already but I won't know until later and maybe I'll talk about it in the next devotional, and maybe I won't.
What I want is rather simple: a few floors of office space at 801 7 Ave SW, where I can put a few of Nvidia's latest devices, with their latest architecture on their latest chips with the fastest interconnects, and power them so that we can have local students access them and learn on them, while also being able to host some sort of show where we transparently broadcast people learning how to use them and talk about it, turning this into a community access center. 801 7 Ave SW has seen a lot of renovations in the last year or so, and from the best of my knowledge is also already the home for a co-location group, YYCX. It is also on the same block and immediately connected to Calgary's +15 as well as a light rail station (known locally as the CTrain or C‑Train) that serves both of our currently existing routes, the Blue Line and the Red Line, so it is central, being downtown, and has great public transit access, importantly to the University of Calgary on the Red Line to the northwest, as well as to Bow Valley College down the Blue Line to the east. Also—did I mention it is downtown?
I did a quick search on Brave but nothing showed up regarding the server co-location group at that address, so some of this information is still out of date, but I am fairly positive this building has great fiber connections municipally within Calgary and should not have power supply issues. I also made a comment on LinkedIn to a member of the task force talking about the broader scope of what Canada needs: sovereign compute that doesn't piss off the locals, doesn't eat us alive with adverse and errant incentives, and isn't owned or operated by foreign companies or other entities that would compromise our interests, and the diversified energy grid, generation, and distribution solutions including a robust grid in order to make sure that this works and can be built upon and does not compromise regular people in this province and nation from being able to access cheap and abundant power. And water, but the water thing is kinda a silly public-sphere consciousness discussion from my current understanding of the real world implementations and engineering challenges. I also mentioned the prairie economic gateway, and supported some other commentators who were talking about compute grants, which I think is in a similar vein to what I am speaking about—something on the scale of at minimum 3 million dollars to stand up something like an NVL72—72 of Nvidia's latest chips, all networked together in a simple system in a rack/tower form factor. We need these, and local workstations, so that locals can get hands-on experience with them if CUDA is a long-term factor that we cannot easily replace in the current contemporary computing and machine learning stack.
Well, that was a bunch, but hey. I think from where I am at right now I am going to try and publish my morning and evening pomodoro entries to the older GitHub Pages blog at sarahlacard.github.io, as I was not able to get sarahlacard.ca up and running in time for the start of October, and I might work towards something a bit more for the mid-day posts, including posts to the old YouTube page, recording them as I create them, and maybe some other things (I'll be avoiding audio and video for now, and sticking to text and computer use as much as possible in terms of modalities, simply for simplicity and safety and also ease of living. I do not like my likeness being remixed or riffed upon, especially by younger individuals or members of the more recent generations) as well as eventually moving up towards something like either streaming it live with an audience and chat, or running a utility that captures my keystrokes in a way that would better allow for training data for a future model—rather than an autoregressive way of trying to emulate or recreate these posts through generative means, I think if I kept this split into 25 minute chunks from more or less writing (including backspaces and other discontinuous actions!) as well as five minutes of editing (upon editing, this is certainly longer, and going to be longer, than 5 minutes to edit these posts), and kept that consistent, maybe, and only maybe, it'll be useful for some reason somewhere or at least be a good thing for me to think about why it either will or will not be, but the backspacing, double checking, correcting thoughts, or phrasing, especially after a long segment of text has been completed or generated is not really a thing in the current models and probably something that is more or less an intermediate step between here and AGI and something that can be done more or less with regular methods currently maybe without having to have any major advances. But really, Twitch is for games, so maybe this is something I work towards re-implementing myself, which is already aligned with plans I had for other reasons, so, hey. All in due course and all things in their own time and times.
I think the most important stuff in modern ML is auxiliary losses and the six major degrees of parallelism. More on that later—maybe. (Certainly not today).
Eight minutes remaining—will I talk about church today? Maybe, maybe not. (Knox Nooners at noon, Wednesday afternoon connections and reflections service at noon thirty, choir practice at seven, meeting with the affirming committee at six fifteen... that's about it. No details for now or today. Dance parties soon hopefully! Let's get more people in the church and celebrating and having fun and interacting and socializing and—dare I say—partying?!) Probably a bit, but more importantly, I suppose is the discussion about the free market and my interpretation of information propagation through such a free market. Really, a lot of my stuff is going to be posted more or less publicly, and more or less online, but really, I am not always going to cross-post things to all platforms, as each location is unique, and as a Canadian person once famously mentioned, the medium is the message. The piece that is missing here is that I am moving towards allowing much more subtle details, both intentional and otherwise, to surface and disseminate naturally through the network, and all interconnected networks, naturally and through an organic progression, allowing the individual who discovered it and whose attention was drawn to it, to seek or capture (hopefully not exploit) the rewards and benefits that it would confer upon them to utilize or incorporate into their own knowledge, undertaking, and communications. To that end, and others, Discord is a bit of a walled garden and cannot be easily accessed by the public or surfaced through a search engine, even Exa, and has a lot of set up required in terms of permissions and access management, which, in my opinion, makes it way less worth the effort compared to other platforms and building or using my own. (LacardLabs CLI soon!) But I'll see what I can make happen on GitHub. And really, I am not fussed; again, a lot of these things are starting to happen on their own and more fluidly.
So, I will talk about some things, but not others, I may or may not exaggerate some things where it will be in my, or our, favour to do so, and rarely ever do the opposite. I will remain charitable in interpretation, and only joke about the outcomes that I want to see in the world and as part of my personal and public life and journey. I may, or may not, avoid things that would singularly single out individuals in my life, regardless of whether or not they have, on the net balance of probabilities, a likelihood of them or anyone else they know encountering the post or writing. I barely even think such thoughts, if at all. Anything else juju that could be improved and better. However, I am not going to stay silent on all topics, and in other situations I may write to almost the sheer exact and opposite effect—I am going through a defamation trial where I am the Plaintiff, and I am learning a lot about that topic, including how far you can push that while essentially getting away with it scott-free or where pursuing any sort of justice is an inordinate amount of effort in both time, research, knowledge, effort, organization, and more, and so where I think things are both factually true, or in the public interest, I feel like I have a rather large leeway in order to simply say what I want, where I want, as loudly and as broadly and as often as I want, regardless of the harm or damage that it may cause. Obviously—or not—this is exaggerated, or maybe it is not, who knows? Whatever that was... (do I have to explain all my jokes and quips?)
I only have a few minutes left, and by a few, I mean barely over two, so let me say this:
(I don't know what I was going to say)
Ah, yes. At some level I was worried about posting publicly again, else I expose my current level or capabilities, or what I am currently doing and the patterns that I utilize in order to make them happen—however, at this point, I am of the belief that anyone who copies my strategies or tactics will fundamentally fail and fall short, even if they are able to gain a temporary advantage by incorporating my approaches and perspectives into theirs, because while it may be asymmetric that they now have mine plus theirs, they do not have the why or how mine exist, nor the backstory or full picture and lore, if you will, regarding how they were acquired, learned, and why and when they are utilized, and what other skills they may have, and if they compete against me, it is almost certain that I will recognize my own patterns, de‑obfuscate them from theirs, and then learn what they have underlying underneath and incorporate them into mine—so with that, and the timer ringing, I will post this to ChatGPT 5 Pro and review it for my own purposes and once‑over editing pass (maybe I'll adjust this workflow in the future and maybe I won't) but this is similar to how we did this back in January so I'll continue with that theme and trend for now.
Sarah
P.S. to GPT 5 - in the future, I'm going to ask you not to respond with code for these devotionals