One Bite at a Time: Swallowing Elephants, Wrangling AI, and Protecting Human Grace

In Ordinary Grace, by William Kent Krueger, Frank’s Dad invites Father Peter Driscoll to dinner. After the meal, Jake and Frank are put to work doing the dishes. After finishing the dishes, Jake asks Frank what he wants him to do.

Frank wakes to the threat of rain the next morning. His parents, Karl Brandt, Sheriff Gregor, and a deputy named Zollee Hauptman are drinking coffee at the kitchen table, with Gregor taking notes. As he listens, Frank learns that Ariel had been with Karl and other friends at a bonfire at Sibley Park before she vanished.

Frank’s mother has often told his father that his friend Gus was crude and vulgar and an influence on the boys that they would all come to regret. His mother and Gus are sitting at the kitchen table and smoking when Frank walks in with his father. Franks tells them they’re looking for Morris Engdahl.

Doyle asks Jake if Karl Brandt confessed to killing his sister. Jake responds by squeezing his face together. Doyle leans down so that his face is close to Jake’s.

In Understanding Deep Learning: Building Machine Learning with PyTorch and Tensorflow, TransformaTech institute writes that large datasets help reduce overfitting. Various sources of data are employed to build robust and efficient deep learning models. Public datasets are openly accessible, often pre-labeled, and have been used to benchmark many state-of-the-art models.

Activation visualization tools can provide insights into which layers are most relevant. They show which parts of an image input contribute most to the decision-making process. Grad-CAM can reveal that deeper layers activate more strongly on tumor regions.

The encoder of an autoencoder captures essential features by applying nonlinear transformations. The bottleneck is the smallest layer. The number of its neurons determines the level of compression. A Wasserstein Generative Adversarial Network reduces instability. The discriminator becoming overly confident can be prevented by adding small random noise to real and fake inputs. Training instability can be clearly illustrated by plotting generator and discriminator loss curves. The discriminator often achieves near-perfect accuracy in distinguishing real from fake samples at the start of GAN training.

Self-supervised learning leverages the structure of the data to create tasks that help models develop robust feature representations without labels. Carefully designed pseudo-tasks that generative models must solve using only available data enable models to extract deeper features. The encoder, decoder, and self attention mechanism design of the transformer model has become the foundation of state-of-the-art models in NLP, computer vision, and more applications. The encoder and decoder blocks are structured to allow parallelization. The self-attention mechanism within the encoder is calculated using query, key, and value matrics. Overfitting occurs when a model captures noise or specific details in the training data rather than general patterns. It is generally observed when the training loss remains low while the validation loss increases sharply during training. Underfitting is evident when both training and validation losses are high. Training and validation loss curves provide insight into how a model is learning.

In The Man Who Broke the Bank at Monte Carlo, the woman lying with her back to Rowbotham is nude as far as he can see. He cranes his neck a little to check on the strewn clothing. Her garments are mingled in with his silk suit, shirt, tie, underpants and socks.

Back in their suite at the Paris after Charles the chauffer picked them up from the Shéhérazade at 5 a.m., Rowbotham tells his Uncle Luigi he would have “snuffed it” for sure. The police had questioned Uncle Luigi, accepting Rowbotham’s assertion that he had fallen asleep. Rowbotham declares on a impulse that he can’t wait until a civilized hour.

In Range: Why Generalists Triumph in a Specialized World, David Epstein shares that the most successful problem solvers spend mental energy figuring out what type of a problem they are facing. Desirable difficulties make knowledge flexible. In 2017, Greg Duncan, Drew Bailey, and colleagues reviewed sixty-seven early childhood education programs. They found a pervasive “fadeout” effect. At ignition, the metal walls that connected to form a joint of the Challenger’s booster pulled apart for a split second. The seventeen flights with no O-ring problems provided no information about how O-rings fail. Burning gas came shooting down the booster heated the O-rings and eroded the rubber surface on five flights. Blow-by, when the rubber ring does not seal the joint at ignition from burning gas, was a life-or-death condition. On Tuesday, April 7, the Utah Stonewall Dems met to endorse our vetted candidates. After thanking everyone who contacted the legislature, spoke up at a committee hearing, attended a protest, our just participated in building community, I introduced our board members. We now have seven members of the board. After announcing our vetted candidates for county races, we gave each of the candidates 30 seconds each to address the caucus, and proceeded to endorse for fifteen of the races. On Thursday, April 9, we had an emergency meeting of the Salt Lake County Democratic Party Platform Committee, because our prior two meetings did not have a quorum present. This time we had a quorum present and forwarded four amendments to the central committee to consider at convention. Friday night, I was still feeling the effects of what had to be a sinus infection in my ear, and decided to try to stay home and tend to it instead of helping set up the convention space the night before.

Bright and early at 7 am on Saturday the 11th, after packing some energy bars and stopping by the gas station to get a couple of soft drinks to add to my lunch box, I arrived at Highland High School to help set up. I said I would help however I could as long as I could avoid breathing on people. I was given some signs to paste on the wall in the auditorium so we could direct candidates to side A or side B.

I then helped tape fundraising flyers to the backs of the chairs in the auditorium. Once that was complete, I headed back to the registration area where my lunchbag, purse, and computer were to grab a quick drink.

I noticed that the people doing checkin were trimming the credentials down one by one as they checked people in so they would fit it the credential holder. After I had them check me in, I grabbed the scissors and proceeded to trim the remaining credentials. At one point, I helped guide people in the line to the checkin tables.

Once that was done, I headed to the stage for convention, and that’s where I spent most of my time. I received a text that a board member of the Stonewall Dems had arrived, and when we broke, I took them to my car to get the popup banner for tabling. Once I got that set up, I headed back to my seat in the auditorium.

Everything was going smoothly until a lot of people started realizing that they did not get their ballot emailed or texted to them. The Salt Lake County Party Executive Director had to manually process every missing ballot claim so that there would not be any duplicate votes. As the line started to stack up, I started asking people what issue they had prior to working with the Executive Director.

The line continued to back up further. I shifted sides to start helping people with the ballots as they were approved. I watched as the line continued to back up, and the congressional candidates started to line up to give their speeches.

I reached out to the chair to let her know that the line would not clear by the time the candidates were scheduled to start speaking. I let the candidates then know that we were going to delay by half an hour. Even though I was scheduled to introduce the candidates, the chair proceeded to do it herself, apologizing later. Though it came as a shock, it was not a big deal, because I could continue to help with people getting their ballots. Once the ballots were collected and the speeches were done, I slipped out the back so I could get to urgent care to have medical care look at my ear. Not surprisingly, they prescribed five days of antibiotics, which is why I went. I knew I was going to need them. In The Enchanted Greenhouse, by Sarah Beth Durst, Terlu asks a winged cat if he has a name. An explorer in a travelogue she once read claimed that there is a distant island with prophetic talking lizards. She asks if there is someone around who feeds the cat.

Terlu blinks open her eyes to see six cakes cooling on racks on a table. Yarrow asks her how many to make. Terlu pats at her hair.

On Tuesday, April 13, I presented my second presentation on responsible Artificial Intelligence for my company. The presentation was titled “AI Better, Safer, Faster: Architecture First.”

It was not an anti‑AI talk. It wasn’t a slowdown talk. It’s a how‑we‑keep-the-speed talk. AI has fundamentally changed how fast we can write code. We need to talk about how we keep that speed without losing control of the systems we own. Before I get into tools, workflows, or Jira, I want you to visualize something with me. Picture a mountain road. It’s narrow. It curves. There are steep drop‑offs. Visibility is limited.

Now imagine driving that road at speed.

Why do mountain roads have guardrails? Not because drivers are bad. Not because engineers don’t trust them. Guardrails exist because even good drivers make small mistakes — fatigue, distraction, misjudgment. Guardrails don’t remove risk. They don’t make the road safe. They make speed survivable.

This is exactly where we are with AI‑assisted development.

AI gives us speed: generation, refactoring, boilerplate, and execution. But speed on a mountain road without guardrails doesn’t make you fast. It makes you fragile. Architecture is our guardrail.

What changed recently isn’t that developers became careless. What changed is friction disappeared. AI introduces new classes of errors—not because it decides poorly, but because its algorithms predict seemingly correct patterns without understanding architectural intent. At the same time, AI removes the delay that used to slow mistakes down. Habits — good or bad — now scale instantly.

I want to share a result that surprised a lot of people. After AI systems decisively beat chess grandmasters, something unexpected happened. In freestyle chess tournaments—where humans could use any tools—they found that the strongest competitors weren’t top grandmasters… and they weren’t pure AI systems either. The winners were hybrid human–AI teams. Not because the humans computed better—but because they used AI strategically, rather than letting it play autonomously. This result matters far beyond chess.

The reason hybrid teams won wasn’t magic. AI excelled at tactics—evaluating positions, calculating moves, spotting local advantages. Humans excelled at strategy—deciding what kind of game they were playing, when to explore, when to consolidate, and when to override suggestions that looked good locally but harmed long-term intent. The winning teams weren’t the strongest humans or the strongest machines. They were people who knew when to ask AI for help—and when not to.

This maps to a broader concept described by David Epstein in his book Range: bounded versus wicked problems. Bounded problems have clear rules, stable goals, and fast feedback. Pattern‑based systems do very well here. Wicked problems are different. Goals shift. Constraints evolve. Feedback is delayed or incomplete. Context matters more than pattern similarity. Pattern‑based systems—including AI—lose accuracy here, because there’s no fixed “correct” pattern to converge on.

This is why software development sits right in the middle. Some parts are bounded: boilerplate, syntax, known refactors, test expansion. AI excels here. But software also contains wicked problems: architectural evolution, domain intent, migration strategy, understanding why something exists. Those require judgment, exploration, and stewardship. Architecture is how we separate these problem types— so AI accelerates bounded work without eroding intent in wicked spaces. This isn’t about limiting AI—it’s about using it where it’s strongest while protecting the parts only humans can own. Think of Architecture as our strategy and AI as our tactics.

There’s a line I keep coming back to:

“AI amplifies good systems… and exposes bad ones.’—StackHubs Global, 2026 [ncsl.org]

AI doesn’t fix architecture. It multiplies whatever architecture you already have.

This is where we need to be intellectually honest. Copilot — and LLMs in general — are not architects. They don’t understand intent, understand historical constraints, or understand why something weird exists. They predict what code is likely to come next.

When we say AI ‘hallucinates,’ what we really mean is this: The model does not know when it is wrong. It will always produce a plausible answer. If context is missing, it fills the gap. That’s not a flaw. That’s a property of how these models work.This is why verification isn’t optional — it’s structural.

And here’s the most important part: AI does not own our code. AI doesn’t carry pager duty. AI doesn’t answer incident calls . AI doesn’t show up to postmortems. We do.

This isn’t philosophical — it’s explicit. GitHub Copilot’s own terms say: “You retain all responsibility for Your Code, including Suggestions you include in Your Code.”— GitHub Copilot Product‑Specific Terms, Section 3 [genspark.ai], [aisera.com]. Ownership never transfers. If it ships, it’s ours — structurally, legally, operationally.

So let me be very clear: This is not a rejection of AI. We are doing the opposite — We’re using AI where it’s strongest: execution, acceleration, and repetition. And we’re pairing it with the thing it cannot replace: architectural intent.

AI‑first development workflows are genuinely impressive. They can read tickets, generate plans, implement changes, and commit code. That’s real leverage.

But AI‑first does not automatically evaluate impact radius, discover prior design intent, or guard against long‑term debt. That’s where we keep assuming guardrails exist — and they don’t.

Technical debt didn’t start with AI, but AI changed the rate at which debt accumulates. Every additional dependency, abstraction, or workaround compounds — especially when generation is cheap. At a certain point, velocity collapses. Not because the team got worse — but because the system became harder to reason about.

When I say upfront design, I’m not talking about big documents or slow gates. I mean knowing the impact area, naming constraints, and giving AI architectural intent. This is lightweight. This is enabling. These are guardrails.

Here’s what “Architecture First” actually looks like in practice. Notice something important: We didn’t remove steps. We added decision points. When reading the ticket, we direct the agent to look for design notes. After implementation, we have it automatically review the implemetation for technical debt, and after committing and pushing the approved changes, it pushes the intent back into the workflow ticket. That’s how guardrails show up in flow.

This is where the promise shows up: less rework, lower cognitive load, and fewer ‘why does this exist?’ moments. And here’s the twist: Good architecture actually makes AI perform better over time — because context improves.

Context is where AI succeeds or fails. Context files act as an external brain in examples like design documents, migration notes, checklists, and long‑running work logs. They’re not overhead. They’re compounding leverage.

I want to pause here and compress everything I’ve doscussed about into a few takeaways you can carry with you. First: AI amplifies architecture — it doesn’t replace it. Whatever structural clarity or mess you have today, AI will scale it. Second: Speed without guardrails doesn’t make you fast — it makes you fragile. Technical debt isn’t caused by AI, but AI removes the friction that used to hide it. Third — and this one matters the most: AI does not own our code. We do. Every architectural decision, every dependency, every shortcut still belongs to us. And finally: Architecture First is how we keep AI making us better, safer, and faster — not just today, but six months from now.

On Thursday, April 26 and Friday, April 27, out team participated in a company-wide AI-Hackathon. As a team, we went after a piece of the architecture that slows down every build. We split the work into three pieces, one for each of the platform developers.

In our 3 million line code base, we have a huge set of java definitions that are used to generate overlay classes that can be used to wrap json in Java. All the overlays were defined in a single module for all domains. We decided to split the module into 97 independent modules, and resolve all the circular dependencies that would be created.

We worked with copilot first to generate the overall plan. Then we split to work on the individual components, using copilot to further refine the plans and generate the code. My piece was the final piece. After a day and a half, we were ready for integration.

Integrating the components was where we ran into problems. Attempting to integrate, each of discovered bugs in our individual components within the generated code. Since I couldn’t get copilot to make the specific fix I was looking for, I had to manually go in and do a search and replace for an area it got wrong in implementation. Most times when I do this, I have to direct copilot to review my changes and update the design documents to prevent it from reverting the changes later.

At 4 pm on Friday I officially went on vacation for the week. The other two members of the team presented in my absence, and I noticed that the Director of Technology recognized BORG (the Build Optimization Repair Group) as doing a great job.

The team was unable to complete in my absence. I was look forward to finishing the optimization when I returned the following week. Activation visualization frameworks were created as a way to see exactly what a neural network fixates on for any classification. For instance, using activation frameworks can be used to isolate cancer cells in an image.

Considering some of these networks are being used to automate credit applications among other uses, visualization should be able to identify the reasons why an individual application was accepted or rejected. It would also help to red flag any bias issues that would require retraining, as well as being able to be used to create a risk mitigation plan or provide feedback to an applicant on what they can do to improve their situation.

We should go even farther. In all interactions where artificial intelligence is used to make decisions about humans, we need a human in the loop to authenticate the factors and the result, and provide feedback to the user on how to improve their situation.

Human verification could be used to help us move from being judgemental in assessments to being actually helpful. Even if another AI tool is used to generate an action plan, the human could verify the action plan and work with the user. The real value in a service is not just how fast in can happen, but how responsive the agents are in responding to the unique needs of people seeking assistance or needing help. This puts a premium on genuine customer service.

There are a number of forces at play, and if left unchecked, they can result in a less than satisfactory system. The first force is the desire to accomplish more with less to meet societal needs. This force to democratize services has with it the temptation to automate as much as possible. Another force in our culture is for corporate culture to cut costs, which they do by automating as much as possible. A third force is artistic individuals and others who do not trust AI at all.

When these forces come together, the direction is for both government services and corporations to replace human workers with artificial intelligence systems. These systems left unchecked will create unchecked errors, and quality in general will go down--with it, the quality of life. Having human support, which is more expensive, would be considered a premium, and as a result, the services would be leveraged by only those who could afford them. Better service would go to the more wealthy in our society, expanding the wealth gap.

These forces are already at work. Companies have been downsizing to shift to AI, but finding that a mistake, as the systems they use to replace people make too many mistakes and really don’t understand people at all. They find themselves killing the goose to lay the golden egg, and having to rebuild the infrastructure that they cost. Just replacing people with AI is the wrong way to go.

The most responsible use of when to use AI is when the product demand is higher than the workforce can achieved, especially when the time is taken up on repetitive tasks in which a human finds very little fulfillment. The proper way to use AI is synergistically—add it to your value chain to help free the human up to do tasks that humans are best at—guiding the systems, validating their results, and working with the customers. This is the true way to democratize the power of AI. With AI in the loop, but not replacing the worker, you can increase productivity and better satisfy the needs of the people who are responsible for the results.

In The Unbroken, by C. L. Clark, Balladarians call a sand storm a flea storm. The storm has become more than just a dark imagining on the horizon’s edge. Sand flays the Qazāli prisoners Touraine and Pruett had caught that morning.

The appraising look that the princess gives Touraine is different in the sunlight streaming through the window. Touraine is suddenly self-conscious of her ragged black cloak. The princess wears trousers and a black coat as well.

Luca sees the way the soldier’s eyes flick immediately to her legs. The soldier sounds like she has been a butler all her life. Luca does not know of a right way to sleep on her legs after twenty years.

When Luca wakes, it takes a long time for her eyes and memory to reveal the cause of her discomfort. She peevishly thinks the rebels could at least have healed the small knife cut in her leg. She feels a pang of guilt for being grateful that the curtains would keep out both the sun and any chance glances.

The nightmare that is cut through by the knock on Luca’s door is a variant on the theme that she’s been experiencing since the bazaar. In the nightmare, if she’s not lucky, fires leap from pyres and devour the city while she runs too slowly to escape. She’s not lucky.

Touraine doubts that Jaghotai and the others will consider any more deals. If she can’t try negotiating one more time, it will take a lot more blood before the rebels get the rain they want. There are few enough to recognize her face as she walks from Old Medina to the Quartier.

A priestess tells Luca that it is a matter of faith. That sharp little word is the only thing Luca struggles with. Back in her bedroom, she sits in one of the comfortable chairs half-reading while she mulls over the question.

Qazāli civilians queue at the Balladairian bank, waiting to be verified to have their grief quantified and paid for. Cantic insists Luca say something before a hanging to encourage loyalty and duty. Luca mutters the words to herself.

In The Faithless, the rains are coming soon. The barracks are now filled with Qazāli, and the fighters are working rifle drills.

Duke Nicolas isn’t as coldly intense as Luca. He tells Touraine that he was led to misunderstand the breadths of Shālan gifts. Taking her hand, he yanks her closer to him.

Aranen din Djasha asks what happened when Luca and Lanquette get Touraine to the ambassador’s suite. Luca follows. After retreating, she paces in the anteroom where Gil and her guards wait.

After many of the nobles have been poisoned, the bells go quiet. Luca’s cloak and coat haven’t been retrieved after the duel she was in. Through the doorway, Touraine sees Aranen leaning over a bed.

When the duke doesn’t answer Touraine’s pounding of the door of the duke’s library and yelling to let her in, she presses her forehead against the locked door. She wonders if Luca’s hunt for Balladairan magic and what they’d found in Champs d’Or got back to him. Taking for granted that Perrot isn’t really her guard is another trapping of the nobility that she stepped into.

Touraine is wearing nothing but a silk chemise when the marquise de Durfort bursts into her chambers. Durfort announces that the death knell is not for the duke. Touraine asks about his supporters.

Later Touraine unbuttons the top of Luca’s chemise. Luca offers to take her bandeau off for Touraine. Touraine’s lips on her send another pulse of heat between Luca’s legs. Luca yanks her shirt and bandeau off simultaneously.

Saturday, April 25, was the Utah State Party Convention. I began the day with the Utah Stonewall Dems Caucus meeting. The atrium where we had the meeting was huge. Wireless microphones and a sound system were provided for our use. Prior to the meeting, I places various queer pioneer laminated sheets on tables around the room.

I began the meeting by thanking everyone for the involvement over the year. Then I announced our vetted and endorsed candidates for each of the races. Afterwards, I took a moment to help connect the caucus members with our legacy. I asked people to name Utah queer pioneers, and as they did so, I took the microphone to them. People that were mentioned included David Nelson, Maude Adams, Jackie Biskupksi, and Derek Kitchen. When each name was presented I asked why they their work was important.

After connecting to our past, I connected everyone to our present. I declared that the meeting was an open forum, and that I would take the microphone from person to person to discuss what was on their mind. I also asked people present to talk to the issues that others brought. We discussed advocacy, connection, and lobbying.

As that was settling down, one caucus member said that he wanted to hear from candidates. So we started hearing from the candidates who were present, giving priority to endorsed candidates, then vetted candidates, and then elected officials before allowing others to speak.

It was fun, as I continued to take the microphone from person to person. As we began to wrap up, I asked the caucus to take note of the people who have put writing that they plan to fight for us. I said, “They’re willing to fight for us. Let’s fight for them.”

After the Stonewall Dems caucus meeting, I attended the Progressive Caucus meeting and the Veterans Caucus meeting. The Veterans caucus was discussing a plan to get candidates to support our issues and to mobilize veterans that supported the candidates that supported the issues the caucus identified.

The remainder of the convention was filled with candidate speeches, and waiting until the ballots were released. Once I submitted my ballot, I headed back home to rest after a long day.

All in all it was a good day. Liban Mohammed made an amazing speech channeling civil rights leaders and applying it to today. He secured his spot in the primary, and faced former congressman Ben McAdams and Senator Nate Blouin in the primary, since they both secured enough signatures to be on the ballot.

According to Rebecca Solnit, in A Paradise Built in Hell: The Extraordinary Communities that Arise in Disaster, if paradises arise in hell, it is because we are free to live and act another way during the suspension of the usual order and the failure of most systems. Stories of the London Blitz; earthquakes in China and Argentina; the Chernobyl nuclear accident; the Chicago heat wave of 1995; the Managua, Nicaragua earthquake, a smallpox epidemic; and a volcanic eruption come with the investigation in depth of five disasters. The book represents an account of rising from the ruins as the ordinary human response to disaster.

In Deep Impact, panic and mob behavior follow a press conference about a six-mile wide meteor headed for Earth. The basic notion is that people’s judgement, social bonds, and humanity are overwhelmed by fear and selfish desire. When Enrico Quarantelli tried to write a master’s thesis on panic in 1954, he couldn’t find any instances of it.

In 1894, the phrase “the scum of Milwaukee” was in the newspapers quite a bit. In 1947, drug companies were less cooperative until locked into New York City Hall and told by Mayor O’Dwyer they were not going anywhere until they produced more smallpox vaccines. In 2005, United States federal officials speculated that a militarily enforced quarantine would be required if a new epidemic were to break out.

The only sign of time on Marisol Hernandez’s face is that she looks of no particular age when she speaks of the moment of rupture in Mexico City on September 19, 1985. In addition to the huge Nuevo León apartment building at Tlaltelolco in the north, the central telephone exchange, the communications tower of the Televisa television station, two large hospitals, several downtown hotels, the buildings housing the ministry of labor, communications, commerce, and the navy, and many large apartment buildings also collapsed. Some people were rescued from the ruins quickly and others after volunteers tunneled for days through unstable heaps with the smell of the deceased becoming terrible in the warm September air.

Three months after Argentines launched an uprising on December 19, 20, and 21 of 2001, the economy collapsed. On New Year’s Eve, the American secretary of state was told there wasn’t an Argentine president. In an economic disaster, while what is wrecked is immaterial and abstract, its consequences are more than tangible. One Buenos Aires inhabitant recalls the feeling of walking back after tens of thousands took over the streets.

In The Starless Sea, by Erin Morgenstern, the son of the fortune teller reaches the age that brings uncertainty as to whether being a fortune teller's sone is something to be proud of. He walks home from school toward an apartment above a shop. He takes a shortcut through an alleyway behind the store.

A merchant once traveled far and wide selling stars. They sold delicate pieces of stars strung on fine chains for wearing around the neck and specimens to display under glass. They carried the stars from place to place in a large sack embroidered with constellations.

Zachary finds himself in a room full of chandeliers. A larger light above is a cluster of glowing globes hung among brass loops and bars. He thinks the contraption may be a model of the universe.

Kat reflects on meeting someone in a bar. “You follow them or you don’t. You open a door or you don’t.” The point in either case is to find out “What happens next.”

In The Pink Wave: Women Running for Office After Trump, Regina M. Matheson and William W. Parsons, tell how the ambition framework assumes personal ambition is a driving force behind stepping up to enter the political arena. In their research, they found many examples of women running after someone approached them , and taking into consideration the political climate of the 2016 presidential election, many ran as a result of their perception of what was occuring in the nation. One respondent stated that one of her friends who was there when she remarked off the cuff “we should all run for office” and served on the recruitment comittee called her to remind her when she said she would run.

A Democratic Party chair asked another respondent to run for office, and she agreed given the current climate. A third respondent stated that she co-founded an Indivisible group following the 2016 election and the group quickly realized that the state had many issues needing immediate attention.

Elder suggests that Republican women interested in running for a state legislative seat would be better moving out of solid red regions. Dittmars examination of Republican women who ran for Congress reveals that many candidates try to break the red-state political ceiling by presenting as more masculine, aggressive, and relentless fighters. The slight incremental growth rate of women serving in state legilslatures may continue well into the future.

In Rears & Vices, by E. M. Caro, Everard contemplates that there is nothing like a summer on Lake Ontario. The ship’s boy has made umpteen rounds of bucket ladles to the Brigitte’s crew audience. Everard hopes the ship’s boy has gotten around to the pirate yet to appear before them.

A moment after he sits on the edge of a four-poster bed, Bellingham St. Clare gasps and scrambles up. D’Arcy jerks his head left to a connecting apartment and makes significant eyes toward Vitaliy. Everard clutches fingers around D’Arcy’s wrist and pulls him toward the door.

Responding to Visaliy’s proposal, Everard comtemplates saying yes. He thinks that “Of course he would” accept and yet he feels he shouldn’t.

In a later scene, the reflected sunlight is very bright around Everard. As the dream fails to fade, he scowls. The least the dream could do was to make the time pass unfelt.

In Turn the Ship Around! A True Story of Turning Followers into Leaders, Louis David Marquet shares that “There are significant benefits to thinking differently about leadership. He’d already been using material from 7 Habits by Steven Covey when Covey expressed a desire to ride a submarine.

While ticking off the accomplishments, Louis said that they had demonstrated key capabilities including a torpedo exercises in the Arabian Gulf, transiting the Strait of Hormuz several times and the Strait of Malacca twice, and a pickup up U.S. Navy SEALs. To Louis’s mind the most impressive statistic was the improvement in the retention results. There were a number of reasons the retention numbers went up, but a key one was that junior enlisted men used to look up to see what the chiefs did in order to get a sense of sticking around long enough to have that job.

In Good Strategy/Bad Strategy, Richard P Rumelt shares how during the Cold War the budget cycle of the military and Congress had created a reactive mindest. Each year, the Joint Chiefs of Staff developed their assessment of the Soviet threat. They had a war strategy, but no plan on how to compete with the Soviet Union over the long term.

In Presence, Senge quotes Srikumar Rao, who says that the process of refinement is a broadcast of intention. No matter how many times New Thought ideas are repeated, they are received with fresh nods of affirmation by many listeners.

Rumelt asserts that believing that rays come out of your head to change the physical world and only thinking of success can make you successful is a form of psychosis and not appropriate as an approach to either management or strategy.

In The Prince Without Sorrow, by Mathree Wilesekara, Ashoka asks Rahil if he thinks killing a deer makes a ruler great. Rahil corrects him, telling him that’s what his father thinks. His father needs a test of his mettle before he is sent off in a few years to govern.

After a meeting of the council, Ashoka’s gold-hilted dagger hits the wooden shelf with a thunk. Sao and he have retired to his quarters to think of a solution for the spirit rampages that doesn’t involve the mayakari. He mimics Arush’s low growl, “Don’t use the mayakari for help. You will do as I say.”

Rahil’s sense of duty is not swayed by his dreams to travel the known world. Ashoka whispers to him that Rahil once told him that he was afraid his position would keep him from leaving the Golden City. Rahil responds that he can’t simply abandon Ashoka.

Arush is more malleable than Shakti had thought. He calls for an impromptu council meeting two days after the dream invasion. Aarya announces that she will not be made to wait. The scholarly part of Ashoka itches to hear more of Nayani’s explanations. The mayakari he has read about do not mirror the ones he has met. Nayani suggests that maybe Ashoka could imagine how Ursa felt when she cursed nature spirits.

Weeks of repressed pain flood Shakti’s consciousness. She realizes it will do no good to risk her life if both she and the mayakari burning in the city square would die if she creates a ruckus. Princess Aarya tells her, “The smell takes a while to get used to.”

Aarya mentions that a mayakari is in the palace. The announcement causes Shakti to still with her sword inches away from the Princess’s throat. Master Kudha has retired briefly to place a healing balm on her injured leg.

At the May 9th meeting, the Utah Stonewall Dems added two more members to our board of directors, bringing our board count up to seven. According to our bylaws, we can have up to fifteen members. Luke Maynes and Logan Howard are our two newest members, with Logan acting as the new recorder.

Mother’s Day has come and passed. I still find myself quite often wanting to just pick the phone and call Mom, and then I realize that’s not an option anymore. It’s been a year since she passed away, and I still miss her.

On Tuesday May 12th, I stopped working early and flew out to meet the team at our corporate offices in Dallas. My adventures started as soon as the cabin lights were turned off for departure.

I reached up to turn on my seat light, and after fumbling a bit, I finally pressed what was the likely button. When the button sounded a chime, and lit up without turning on the light, I tried to turn it off and heard another chime. The steward’s voice then came over the cabin announcing to only call if this was truly an emergency, and that they would answer as soon as we were safely in the air.

I sat there truly embarassed, and cracked the window shade open so I could have enough light to read by until we were safely airborne and the steward approached my seat. I explained that I was just trying to turn on the light, and she asked if she could help. When I nodded my assent, she showed me a tiny oblong button just aft of the lamp.

As she walked away from me, she approached another passenger on the other side of the plane three seats behind me. Watching her turn on that passenger’s light, I breathed a sigh of relief, realizing I was not alone in my confusion.

The rest of the flight was fairly noneventful. We landed at DFW and I called my Lyft. The driver was polite, and everything was fine until the driver dropped me off at the entrance to my hotel.

I walked up the steps to the hotel entrance to find the door to the Warwick Melrose was locked. I noticed the buzzer button on the side and gave it a push. I then waited a moment with no response.

Looking through the glass doors, I noticed a couple of men having a conversation at the check-in counter. So I pressed the button again and then waited.

Eventually the clerk came to the door, and said, “Sorry about that. I was with another customer.” I told him that I had a reservation, and he responded, “Let’s check you in,” and he proceeded to look up my reservation.

 As he worked, I noticed he had a somewhat delicate touch, with an almost feminine manner. I talked about this trip being a precursor to a “significant” undefined promotion after the upcoming merger and about my anxiety about trying to make the right impression. He then handed me the room card, and said, “Here you go. I updated you to the deluxe suite for free.”

Once I got to the door leading to my room, I noticed that the card wasn’t working. I tried several times, and when I was about to give up, I noticed I had the wrong room. Padding down the hall, I tried again, and the card key worked.

It was a challenge to find the light switches and the charging ports, but once the room was lit up, I was impressed. It had a living room and a kitchenette with a separate door. There was a TV in both the living room and the bedroom.

As I tried to deduce why I had the upgrade, it dawned on me that the clerk was probably gay. And then I realized that I was in Oak Lawn, the gay district of Dallas, and I was among the LGBTQ+ community. We tend to take care of our own.

Early the next morning, I packed everything up, snacked on my trail mix from the flight, and grabbed a cup of coffee. I did some quick research to find that I could buy a breakfast sandwich on the ground floor where my company was located.

Downstairs, I checked out with the night clerk and his manager, a young black man who also appeared to likely be gay. I thanked them for the room, and he said, of course, and told his manager that I was getting a promotion. I corrected him saying it wasn’t done yet, and he insisted that he was confident I was going to get that promotion.

I then mentioned that I had belatedly realized that I had chosed a hotel in the heart of Dallas’s gay district, and it was good to be among community.

I thought to myself that the office shouldn’t be that far away, and asked if they had a bus stop nearby. They told me it was just across the street.

I headed out of the hotel, looking for the bus stop, and crossed the road to the other side. I pulled out the map app on my phone and got the public transportation directions. It guided me up the road to where the bus stop was supposed to be, and I crossed the road to look for it. I even approached the businesses to see if any were open so I could ask directions.

There were no signs at the location saying it was a bus stop, so I crossed the road again, and when I looked back, there was the sign. It was a small sign, visible to the traffic if they looked for it, but not at all visible from the sidewalk. Once I had located it, I crossed once more and hung out under the sign.

Fortunately for me the bus was running a little late, so I didn’t miss it. According to the directions, I had to ride the bus for nine stops, then get out and walk to the next stop.

Once I settled on the bus, I noticed that the bus would not stop if no one requested a stop and there was no one there. I prepared myself to reach over and press the strip on the wall of the bus when my stop was next, monitoring my app so that I knew when it was time.

The sixth stop was the West Transfer Station, and the driver announced that she had to take a break to use the bathroom. When she got back, hoping she would help me remember, I told her I needed to disembark at the third stop coming up.

About 10 minutes passed, and when someone asked when she was leaving, she announced it would be at 8:05 am. At that point, I realized that I would miss my connection, and not get into the office until about 8:35 am. I decided to disembark, and walk six minutes to the connection.

Once I got to the connection point, I realized I was too late again, and to get to the office, I needed to walk another six minutes to another stop and catch the 264.

I made it to the station in plenty of time. There were a number of people hanging around and I sat on the concrete bench while I waited. A bus pulled up, and I realized it wasn’t the one I was looking for. A couple more buses stopped, and then I finally saw it off in the distance.

The number 264 was as clear as day on the black digital marquis of the bus a block away at the traffic light. I got up and stood next to the people standing at the bus stop. The bus approached the stop.

And then it flew by at 30 mph, not even slowing down. I turned to the young black woman next to me and said incredulously, “That was my bus.” She replied, “You need to stand under the sign. They won't stop if you're not under the sign.” Nice to know.

Now I knew the rule, and knew that the bus drivers have an attitude. There were definitely signs of differential treatment, with the hard to find signs in the LGBTQ+ community, to riders being left in the dust in the black community. This was another reminder of the discrimination that still goes on in subtle ways in Texas.

At that point, I gave up on public transit for the day, and called a Lyft. It was another six dollars, but they got me to the office building with plenty of time.

Inside, I went to the shop mentioned by Google AI, and discovered that they didn’t have breakfast sandwiches. It’s hard to believe I got snookered by an LLM hallucination again. Anyway, I had a lovely chat with the owner operator, and a delicious blueberry parfait with a mocha while I waited.

I gave up waiting for my coworker to show up and lead me up at about 8:05 am. I approached the security desk, signed in, and said I needed to go up to my company’s headquarters. As I signed the log, I noticed that I wasn’t the first that morning, and the security team called the elevator for me, telling me which elevator in the bank was programmed to take me to my destination.

I stepped out of the elevator at corporate headquarters and saw the wide, open stairs that led up to the next level. My lead and program manager found me in the lobby, and we navigated to the conference room where we were to meet.

As I set up my laptop for the presentation, we discovered that I couldn’t extend my notebook computer to the overhead screen. Since my narrative was in the notes of each slide, I shared my powerpoint presentation with the program manager so he could page through it while I read and ad-libbed to my narrative.

While we were waiting, my lead made me guess who each person was coming into the room. These were all people I worked with remotely, so I had no clue, and felt really terrible. Then a very familar, deep soothing voice comes up next to me and says, “Guess who?” I knew immediately who it was, but for the life of me, his name would not come to mind even though I work almost constantly with him online.

He looked a lot different than his profile in Teams. In fact almost everyone did. And yet, I felt even worse, because my memory betrayed me. When he finally said his name, I said I didn’t know why I was drawing a blank. I very well knew who he was, but the name just wouldn’t come. When he said, "I'm Lewis," and oh-duh went through my head, flabbergasted at my failure to recall his name.

Another individual came in that looked familiar, and while he was there I couldn’t place who he was. It dawned on me later before I left the building that he was the Senior Vice President of Technology, the lead to my skip lead, and I knew his name immediately.

Once we had practically every backend developer in the company either online or in the room, I ran through my presentation, with encouragement from the directors, leads, and VPs in the room. We discussed primarily the Java system upgrade, why we doing it, the AI integration I was using to speed up the process, and how we planned to have developers or their leads merge their changes into the branch. We also discussed what the benefits of the new system were—mostly perfomance and scalability, and then I opened it up to questions and concerns. Probably the biggest concern was how they were going to able to run two builds simultaneously. We discussed the possibility of adding a second virtual environment or adding more capabaility to the existing virtual environment.

Once we addressed all the concerns, we turned to the topic I really wanted to bring up, which was what kinds of issues were they facing co-developing with AI. A couple of issues I have seen is that developers working on a shared project still get integration issues because the working context is different for each developer despite creating a shared context, and the agent lapsing into what I call “dementia,” when the context suddenly drops and the agent starts back up again as if everything you just did doesn’t exist.

After a couple more meetings, it was time to head out of the office and catch my flight back to Salt Lake City. I rode the elevator down with my coworkers and placed an order for my Lyft on my cell phone. The elevator let them out at the garage and I continued my way down to the lobby and out the main door.

Once outside, I realized I left my carry-on in the conference room and headed back in the building, telling the security desk I left my bag upstairs, and proceeding to cancel my Lyft on the way up.

It didn’t take me long to get my bag and get back to the elevator. Breathing a sigh of relief, I looked down at the app to realize I hadn’t finished canceling the ride, so I aborted the cancellation, and headed out the door to the pickup spot.

I noticed the street address seemed weird, but not knowing the block I was on, I assumed it had to be one the streets adjacent to the building. As it was getting hot, I took off my sweater and stored it in my carry-on. I then sat to wait for the Lyft, watching the app for the driver.

Eventually, I got a notification that my driver was there, but he was nowhere to be seen. I was about to call, when he called me. After a couple minutes trying to give him the correct address, we decided to cancel the request. I asked him to not treat it as a no show, which would charge me, because the application has messed up.

Once I canceled the request, I tried again. This time it got the right pickup location. When I went to verify the address it used for the airport, I discovered the drop off location was actually a three bedroom home near Dallas Love Field, not the airport itself. I quickly changed the destination address to the airport on Herb Keller Boulevard, and then waited for my driver.

 The ride to the airport was pretty noneventful and the driver was very polite. As we made it there, I realized I had only about an hour and a half until my scheduled flight. Prepping to go through security, I ate what I could of my trail mix, tossed the rest, and left my unopened bottle of water where someone could grab it. I then proceeded to dig in my purse for my drivers license.

It was gone! I searched every compartment in my purse, my backpack, and my carry-on three times. It simply wasn’t found. I immediately recalled the story of a friend of mine who had been stuck at their destination because the same thing had happened to them.

Calming down, I tried contacting the Utah DLD chat line. Once I had someone online, they told me that they were forbidden from emailing duplicate drivers licenses, and that I should speak with the TSA information at the airport.

In front of me were the long lines of people going through security. I didn’t see a TSA information booth. Looking up, I saw a general information booth a hundred feet to my left, so I headed over there, hoping they could guide me to the TSA offices. The directed me to turn around and try the corridor to my left.

I took the corridor to the left, which was immediately before the checkpoint. Walking down the hallway, I only noticed one person, who was cleaning the rooms. When I asked them where TSA information was, they said it was on the other side, and that I would have to go around.

So I turned around and turned left at the end of the corridor and walked past the checkpoint, which had to be another hundred feet, and saw no corridor on the other side. Not knowing what to do, I turned around and walked back.

On may way back, right before I reached the wall leading to the corridor again, a gentleman wearing one of the security uniforms asked if he could help me. I told him my situation, and he told me that he was one of the supervisors. Then he told me what I needed to do. I needed to enter the checkpoint queue, scan the QR code on the sign for people that don’t have “real ids,” fill out the online form, pay the sight $45, get back in line and show them receipt. At that point they would call a supervisor.

As I walked down the queue, there were only three people ahead of me at the ID checkpoint. As I walked toward the red sign, the security personnel, said, “Over here.” I responded that I had to go to the sign first. When I was done with the online application, there were now six people ahead of me.

Once I finally got to the front, I showed her the receipt, telling her that I had lost my drivers license. She directed me to my right to an unopened scanner, and said “Wait here. I need to call a supervisor.” As I waited, eventually two gentlemen sauntered over from different directions, one of them suddenly taking a right and heading another direction. The remaining officer looked around, saw me, and asked, “Is it you?”

Once I confirmed that it was me, he took me back to a corner to answer proof of identity questions, like what organizations I have worked with and if I knew my daughter by her name and when she was born. Having satisfied his questions, he had me sign a form, and I was able to finish going through security.

As I was taking my two laptops out of my backpack, the screener told me that the computers stay in the bags. Sure enough, my backpack got derailed for the manual inspection, but it wasn’t the computers. The inspector dug in my backpack and dug out a notepad, They said that the had to check just to make sure it wasn’t explosives.

Once I got everything back together, it was about 45 minutes until the flight’s scheduled departure, so walked as quickly as I could across the airport to the gate. As I got there, I discovered the plane hadn’t arrived yet and the flight was delayed by 30 minutes. Realizing that I had a one hour layover, I approached the flight attendant’s desk saying that I might miss my connection. She checked and told me that the connecting flight was delayed as well. I returned to my seat anxious that this was going to be close.

Once we touched down in Phoenix and I deplaned, I walked as quickly as I could to the connecting terminal. I had no idea that a single terminal at the Phoenix airport could be so long. I use the people mover sidewalks as much as possible, and when I was stuck behind a family on one , I took long, fast strides beside the next mover to try to get around them.

Eventually I arrived to find the flight was delayed not only the original 30 minutes, but another 15 minutes for people arriving on connecting flights. These gave me a new appreciation for Delta Air Lines, who care enough to slow down to care for their passengers.

The rest of the flight was noneventful, and eventually I arrived home just after midight early Friday morning. So much for trying to arrive on the same day.

The following week, in preparation for the corporate merge, we had a two day CI/CD transformation workshop, where my lead and I were there to represent the necessary changes to the backend structure to allow faster builds and deploys. The Vice President of Technology invited me to the workshop because she felt that I was solution-focused. It was also interesting when she kept calling both my lead and I “leads,” because I don’t have any direct reports. I wondered what she knew about my future with the company that I didn’t know.

On Monday, June 1, my company merged with Auctane to becomeShipStation Global. The CEO made the announcement, and that they would start having conversations about “soft landings” with people who were in a redundant position.

That Thursday the announcements started coming through the back channel. The first name that I heard that was gone was our program manager. While I was talking with a technology director, we tried bringing my lead in for an architectural decision as part of the Java upgrade. He didn’t respond.

After that meeting, my skip lead wanted to know if I was free. Nervously, I replied that I was. In the one-on-one he told me that my lead and the other person on the team had been impacted by the reduction in force. I would be reporting directly to him, the Senior Vice President of Solution Architecture. That meant that I would pretty much own the core systems, and would potentially work with the other architects that report to him. I asked about collaboration channels, and he said that he would arrange that.

By the end of the day, I had survived. The survival itself was bittersweet. I miss the banter that we had fallen into as co-workers. My days are too silent now. The company is promising no more waves, but I am not sure that I can trust that. The company we are merging with had a very public reduction in force 2 years ago, and I don’t see what’s going to stop that again.

In some ways, I feel like my pending promotion may have made my lead redundant. My value was realized because of all the things I have been working on. Ever since I was promoted to architect 2 and laid off in 2009, I have been guided by a principle to be indispensable and not redundant. It may have saved me this time.

In System Design Interview, Alex Xu outlines a web crawler design. Web crawlers begin by collecting a few web pages and then following the links to collect new content. In search engine indexing, they collect web pages to create a local index for search engines. Web archiving is the process of collecting web information to preserve the data for future use. An example of web mining is top financial firms using crawlers to download shareholder meetings and annual reports to learn key company objectives. In web monitoring, crawlers help monitor copyright and trademark infringements.

Another interview question could be to design a search autocomplete system. The first step is ask enough questions to clarify the requirements. In the example, the candidate asks if matching only supported at beginning of a query, how many autocomplete suggestions the system should return, and how the system knows which suggestions to return.

In Greenteeth, by Molly O’Neill, light changes again as Jenny, Temperance, and Brackus near the bottom of a mountain. Temperance, panting, asks if they are back. Brackus twists around and says, “Aye.”

Jenny groans, asking if they have to walk all the way back. Brackus says he has the ability to cut a new opening to the roads. Jenny scowls, asking why they spent so much time looking for other openings.

In The Last Days of the Romanovs: Tragedy at Ekaterinburg, Helen Rappaport tells how the Tsar and Tsaritsa feel convinced that fate will intervene. Fearful of lynching, the Urals military comissar decides to send the train to Ekaterinburg’s freight station No. 2 at Sartash. There the Ural Regional Soviet receives the Tsar and his family into their hands for “detention and surveilance.”

To Vicky, Queen Victoria’s eldest daughter, Alexandra was burdened by the enduring sin of pride. Ingrained in Alexandra was duty to family and to the state. It took Tsarevich Nicholas five years of waiting a superhuman persistence to wear down Alix’s resistance to conversion from Lutheranism to Russian Orthodoxy so their marriage could take place.

For Dr Evgeny Sergeevich Botkin, time in the Ipatiev House drags terribly. In early July, the doctor begins to intermittently write a letter. He doesn’t think his is fated to write again from anywhere. He knows he cannot share his musings with those around him. Tall and stout, he has a neat beard, gold-rimmed spectacles, and dresses formally even in confinement.

In The Kitchen House, by Kathleen Grissom. the captain returns home just long enough to turn the place upside down. The sick child he gave Belle can’t keep her food down during the day. When Belle was little, living at the big house, the captain would come riding up the back way on horseback while she was waiting for his carriage at the front door.

Marshall doesn’t respond to Lavinia’s engagement any better than Meg. He states that his still has rights to Lavinia, whose indentureship is not well-defined.

In Her Soul to Take, by Harley Laroux, Rae lets water cascade over her face and through her hair, hoping it will wash away her arousal. It doesn’t work. During the walk home after class, the rain had stopped, and even though the clouds had cleared to let through the a little moonlight, the darkness of the forest was impenetrable.

Rae keeps expecting Leon to pop up and make good on his filfthy threats. He managed to stop her from getting her video ready to upload. Rae is determined that his dirty-talk distraction won’t convince her to go along with his bargain.

The scent of pines, the rattle of aspen leaves, and the smell of rich, damp earth take Rae back to her childhood. On the property behind the cabin is a spot where a massive pine had fallen years before she was born. Leon's moods range from dangerously quiet calm to raging impending apocalypse.

On June 12, Brooke, Andy, and I watched a live perfomance of Xanadu at the West Valley Performing Arts Center. Midway up, the seats had a perfect view of ground stage and the elevated stages all around the theater. A muse finding herself in Venice Beach, California, is tasked with inspiring a local artist who wants to open a roller disco without revealing herself as a muse, falling in love with a human, or creating art herself. Of course, the jealous muse who wants the reward of Xanadu conspires to make all three happen.

My various roles are starting to consume more time. I have started attending Development Committee meetings and Strategic Planning meetings for the ACLU of Utah, which will likely be another three meetings a month on top of the other meetings I already have.

Good Trouble Organizing has added me to their planning meetings, and that means another two meetings per month as well. Add in the monthly House District Chair, County Caucus Leadership Meeting, Salt Lake County Party Executive Committee meetings, Utah Stonewall Dems meeting, and Veterans Committee meeting, you start to get the picture.

My response was to delegate. At our Utah Stonewall Dems Board meeting, after our monthly reports, I asked Jared to take over Queer Connect, and asked Jahn to submit me a list of what topics he would like in the newsletter. I would then craft the newsletter, send it for review, and then distribute it to the mailing list. That frees me up from having to do research about what is happening in the communities, and potentially catch up on my chore list. I even finally got my tub cleaned yesterday.

Dem Bones is an outdoor summer event that the Utah Stonewall Dems puts on each summer. It is a cummunity potluck, and an opportunity to gather for the LGBTQ+ community. We provide the drinks and the ribs. It is also a chance for candidates and elected officials to speak, and to endorse vetted candidates. Any candidate who has been vetted by the board in which there is not an endorsed candidate currently running is eligible for our endorsement.

My company is in the process of restructuring after the merger. WwexGroup and Auctane have now been joined to create SpeedshipGlobal Worldwide. I have been focused on getting the 3 million lines of code in the backend ready for delivering solutions faster.

I have spent the last few months upgrading our system from JDK 17 to JDK 25, and am nearing completion. All I have left is getting one field to show up in the logs, merge the latest release, and shepherd it through testing into production. Already, we are seeing 100% throughput improvement.

In parallel, I have been working on modularizing the codebase by domain. This is necessary to make releases faster. With the intial breakout, developers will be able to see overlay definition modules closer to the features that use them, changing an overlay won’t trigger a full build locally, releasing by domain becomes possible, the deployed memory footprints become smaller, and domains will be easier to move to their own repository.

I’m swallowing the elephant one bite at a time. I am in the process of extracting the second domain and getting the compilation to work. I am using AI as an assistant to help generate helper scripts and fix bug patterns that it can reliably fix faster than me. The rest is on me.

You can’t just rely on copilot. It can be amazingly accurate at times, but other times it goes off into the weeds, or starts making the wrong changes to fix the tests that it broke. It can be very difficult to wrangle it in. I find it necessary to watch over it’s shoulder, and am not afraid to hit stop when it tries to do a full scan that is not required or seems lost. At that point, I can attempt to guide it, or just take over. Part of the new calculus is learning when it’s faster and more reliable to use the tool, or when to do it yourself.

Two weeks ago, the VP of Solution Architecture informed me that I was going to be moved to the Invoicing pod. Last week I found out that the SVP of Technology ruled out the VP of Technology, and said that they needed me to stay under him, and they are assigning an architect to the team instead. My new responsibilities will be investigating frontier technology to integrate the disparate data and helping analyze it for growth opportunities, once I complete or handoff my current projects.

This is kind of funny, because it brings me back full circle. Twenty years ago, my computer science and engineering thesis for my master of science degree was in graph-based data mining, technology that lived at the crux of data science and artificial intelligence. It will be good to be back at the forefront of technology.

Just a three days ago, I got a call from my friend Deborah. A voicemail, actually. She said that she had a car accident and now her car was gone. This is a hit for someone who just recently lost her wife. New to cellphones, she wasn’t responding to my calls or text messages, so I drove out to her place to check on her. When I asked her if the police offered her a ride home, she said she walked back because she was crossdressed. I should know how that it is.

I was floored. Yes, there was a time when we should fear the local police as a trans individual, but she shouldn’t have felt she needed to do that. It’s a matter of simple courtesy. Most police officers in this area are better educated that they were 20 years ago. My friend, who is in her 70s should have been offered a courtesy ride at the very least.

Deborah is struggling to use her new phone. It’s still a frustrating technology for her. Even in today’s well-connected age, it becomes easy for people to become cutoff from systems of support when the technology is not intuitive for older generations. Even I struggled when I upgraded my phone, until I learned all the swipes and cryptic buttons I had to push. When we design products, we need to remember to try not to be too clever by default.

I also got to celebrate a couple of financial milestones this week. The Traverse is paid off and I reached 17% equity in my home. Of course, I started getting harassing calls from the credit union because there was 55 cents in fees on my balance that was not included in the calculated amount for the final payment.

I started getting emails and texts claiming that I had missed a payment and that they would start charging late fees. Because I have multiple loans through the credit union, I went to my account to see if maybe I had accidentally missed one of them. They were all made on time, and they matched the amount on the statement.

I called the customer support line, and once they dug into the account, they saw I had a small balance on the account remaining, and to them it made sense that I would get the notification. It took me a long time to argue that I had not missed a payment, that the next statement wasn’t due until July 1, and that the messages were harassment. The fee is for insurance that covers the loan if I should die or become disabled. There was no loan amount left! The fee was ridiculous, and if it did add to the balance, I could pay it off the following month.

I was told that the interest was a daily accrual, and that I could just check the website for the payoff amount. That is nuts. If the statement says a final payoff of a certain amount by a certain date, it should have the anticipated included. It’s ridiculous to think that I should go to their website to determine exactly how much to pay, and it’s doubly ridiculous to get harassing notifications to falsely accuse me of not making a payment I had actually made.

The customer rep transferred me to the account rep, where I had to make the same arguments once again. They just don’t get it. She waived the 55 cent fee and assured me the harassing notices would stop. They didn’t.

I got another text yesterday, and after threatening to sue for harassment, I sent the one word reply, “stop.”

Ultimately, whether we are evaluating a fractured O-ring, modularizing a monolithic 3-million-line codebase, or fighting a credit union over a ridiculous 55-cent daily-accrual fee, we are constantly forced to confront the exact nature of the problems before us. Life rarely hands us clean, pre-labeled datasets or simple, bounded paths. Instead, it throws us into the "wicked" realities of unpredicted bottlenecks, sudden procedural shifts, and the cold, automated text alerts of rigid institutional systems. We see this mirrored in the tragic, confinement-weary worlds of The Last Days of the Romanovs or The Kitchen House, and we live it when a company reduction in force abruptly turns a bustling collaborative workspace into a silent room where we must carry the core architecture alone. In these heavy, bittersweet moments—just like when our AI coding assistants lapse into "dementia" or make wrong changes to fix the very tests they broke—blind velocity doesn't save us. True resilience requires us to swallow the elephant one bite at a time. It demands that we watch over the tool’s shoulder, refuse to be seduced by cleverness for cleverness's sake, and remain entirely unafraid to hit the stop button to take back control of our own design.

As we see in the grass-roots solidarity of the Utah political caucuses, our upcoming Dem Bones potlucks, and the spontaneous support networks of the LGBTQ+ community in Oak Lawn, humans are fundamentally designed to take care of their own. Powerful stewardship means recognizing that leadership is not about managing direct reports; it is about extending a hand to bridge the widening gaps in our fractured world. It is the simple courtesy of driving out to check on a friend like Deborah when intuitive technology cuts off older generations from their support systems, and it is the fierce advocacy required to ensure that a trans elder never feels forced to walk home alone from a car accident out of fear. Leaving our systems unchecked—whether in software deployment, corporate restructuring, or civic governance—does not make us fast; it makes us fragile. AI can scale tactical execution at thirty miles per hour, but like a bus flying past a waiting passenger, it will entirely miss the human context if we aren't standing precisely "under the sign." Technology cannot hold architectural intent, it cannot navigate the nuances of grief or community legacy, and it absolutely cannot step up to answer an incident call or carry us through a layout.

Like Zachary exploring the chandelier-lit universe of The Starless Sea, we are always standing in front of unmapped spaces where the future is entirely unwritten. We follow the call or we don't; we open the door or we don't. As we step into our own full circles—returning to the cutting edge of frontier technology, graph-based data mining, and massive scale optimizations—let us move forward with intellectual honesty. Let us delegate our burdens, clean our tubs, and leverage machines to accelerate the repetitive boilerplate of our daily lives, while fiercely guarding the strategy, the validation, and the deep, empathetic human care that can only be delivered when a human is proudly, undeniably, and defiantly in the loop.