Modern Problem Solving

Shifting my X to track research I do into different opinions in bio, politics, health, and consciousness.

[DataRepublican (small r) @DataRepublican]

Every Republican legislature in America should immediately pass legislation to:

1) Ban sanctuary cities
2) Ban Ranked Choice Voting
3) Ban ballot harvesting
4) Require citizenship to vote
5) Require photo ID to vote
6) Ban the National Popular Vote Interstate Compact
7) Ban foreign entities from buying property
8) Suspend any gas taxes
9) Institute school choice
10) Abolish property taxes
11) Open non-juror records to public inspection
12) Mandate secretaries of state to share complete voter registration rolls with DOJ
13) Mandate voter roll list maintenance
14) Offer rewards for convictions involving fraudulent use of Medicare, Medicaid, SNAP, Social Security, or other entitlements

Republican governors would be wise to use this time to convene special sessions.

[Cynical Publius @CynicalPublius]

I'm seeing an increasing groundswell on the Right saying the entire election was stolen.

I'm also seeing analysts I respect saying that the voting shift was consistently demographic-based across the board, a trend which would not be possible in some sort of electronically manipulated election.

I'm not saying voter fraud does not exist. It does. Alaska is an easy example.

I only ask that you remember that the Color Revolution plotters expected to win the midterms, and expected to use Trump and the MAGA movement contesting the results as a "redline" that could be used to declare Trump "illegitimate" and overthrow him.

That "redline" could occur anytime between now and January 3rd (the date the new Congress is seated).

I'm not telling you not to look for voter fraud.

I AM telling you to use good judgment and discernment and to realize that there are bad actors who may be stirring up some of these arguments as false flag ops.

Please be discerning.

[Ariana Thacker @m0ldilocks]

In 2005, Dr. Ritchie Shoemaker published Mold Warriors. In it, he pointed out a pattern that mainstream medicine largely overlooked: a connection between mold exposure, elevated anticardiolipin antibodies, and miscarriage risk.

Twenty-one years later, the clinical data is catching up.
A July 2026 study evaluated pregnant women living in mold-and mycotoxin-contaminated homes. The findings are staggering:

The Numbers:
- 90% Complication Rate: Out of 21 pregnancies across 8 women, 19 resulted in serious complications.
- 38% Miscarriage Rate: Over a third of pregnancies ended in miscarriage.
- 33% Premature Labor: One-third delivered preterm.

What Was Observed
- Placental Damage: Calcification, chronic villitis, placental infarcts, and gritty membranitis.
- Congenital & Birth Defects (38% of infants): Cardiac anomalies (levocardia, patent foramen ovale, VSD), renal hypertrophy, cranial nerve issues (ptosis), and distinctive gray/black skin discoloration.
- Postnatal & Lactation Risks: Oropharyngeal dysphagia, reflux, respiratory arrest, and apnea during feeding while living in contaminated environments.
- Primary Pathogens Documented: Aspergillus/Penicillium in 100% of cases, Stachybotrys in 75%, and Chaetomium in 62%.

Why This Matters
For decades, patients and forward-thinking clinicians have connected environmental toxicity with reproductive heartbreak-often without the formal research backing to prove it in conventional settings.

The paper’s conclusion is unambiguous: in-utero exposure to indoor mycotoxins and Stachybotrys correlates strongly with adverse gestational, placental, and neonatal complications.

It shouldn't take two decades for clinical validation to reach standard care.

At MoldCo, this is why we don't treat mold exposure as "just allergies." It is a multi-system, multi-generational environmental illness. Research like this brings us one step closer to the standard of care these families deserve.

[Modern Problem Solving @ModProblemSolv]

Looks to me like a good pivot in Republican communications strategy would be to have all speechs based on immigation, fraud, Iran, and crime just be composed using excerpts from actual speechs done by the Clintons, Biden, and Obama from the last 25 years. Hindsight is 20/20 but it occurs to me this is what Trump should have done to show the positions he took were pretty actually moderate 10 years ago and therefore should now have been reasonably Bipartisan.

[Modern Problem Solving @ModProblemSolv]

@elonmusk as a burner you'll know one of the magical parts of the experience has been it forces people to unplug and be present. This puts people in a more open mood making the playa such a magical experience. Please consider making a change to @Starlink policy that blocks access on the playa from social media sites or provides some other throttling or blocking mechanism. I can't imagine it meaningfully impacting revenue but it will have a transformational experience on the event.

[Modern Problem Solving @ModProblemSolv]

@_welf @daylightco @supermirror_app Thank you for building this. I was hours into designing something way worse then daylight customer service told me this existed. I could not more stronly recommend buying it

[Ariana Thacker @m0ldilocks]

Three years ago I got sick from something most people overlook: mold exposure.

Today, that journey comes full circle: we’re announcing MoldCo’s $8M Seed (total $11M), led by @cantos + @collabfund, to build the new standard of care for mold and chronic inflammation.

The mission is clear: to serve patients who’ve waited far too long for answers.

To celebrate, we’re giving away 50 free lab tests. Comment “labs” to claim one and we’ll DM you!

[Modern Problem Solving @ModProblemSolv]

@tomcrawshaw01 Leak

[Alex Veremeyenko @alex_verem]

Grok 4 + Emergent is basically cheating.

I used the smartest AI model inside the fastest no-code builder.

The result? Unfair.

Here are 4 wild builds we made with this combo 👇

[Modern Problem Solving @ModProblemSolv]

In a recent meeting with our product team, I discussed the concept of designing for AI in SaaS products, something I call AI Primacy. I was surprised when members of the team expressed skepticism that AI could automate the majority of use cases in real estate, accounting, and back-office maintenance coordination.

I was surprised by these positions, but they were more widespread than I expected, so I captured the pro-AI position in the following document to share more broadly. For most of the people I am connected with here, this should be immediately clear and not require review, but I am sharing it because it has been helpful with skeptics and "tech civilians" who don't fully grasp the scope of what is happening.

To make the point, I build on an analogy from @chamath comparing AI to the Apollo program, and I also share concepts about the differences in machine intelligence and why customer service is not a moat against AI.
Anyway, I'm sharing the document here for others to use (if needed).

Designing for AI Primacy: The End of Human Execution in Product Design

Background
AI is on track to replace most white-collar support labor in categories like customer service and property management. This is because to-date, product design has assumed two independent parties: one with a need, and another using tools and policies to meet this need. That second party is typically a human who applies process knowledge, follows defined workflows, and communicates outcomes. With advances in AI, meeting this need with a human is quickly becoming unnecessary.

I frequently hear skepticism around how quickly this shift will happen. The goal of this piece is to lay out why that skepticism is misplaced, why the technical hurdles are mostly behind us, and why continuing to build for human agents is quickly becoming a legacy assumption.

Moving to AI Primacy
Complex human-managed workflows involve a mix of tasks: retrieving information, discerning intent, managing exceptions, summarizing context, and taking action. In conversations about the future of our product, I continue to hear a cautious posture toward AI’s ability to manage these actions. For example, someone voiced concern that AI might not be ready to handle lease negotiation because it requires policy interpretation, human tone, and exception handling. Another shared a frustrating experience with a hotel check-in bot that failed to handle even a modest deviation from the standard workflow.
These concerns are familiar: What if the policy is unclear? What if the resident pushes back? What if the model misinterprets tone or intent? This skepticism is common because people are still hesitant to move toward a world where the default assumption is that AI can primarily manage these activities. While understandable, it is wrong because well-designed prompt engineering, supported by strong knowledge resources, and multi-step reasoning (when needed), can already manage most of the problems I've faced in my career.
As such, the fact that models cannot yet handle every edge case should not delay us from shifting our design paradigm to rely on AI as the default executor. This is because the scale of investment in this space, across both capital and human intelligence, is unprecedented over the past century and will close the remaining gaps in capabilities faster than we can roll solutions into the economy.

Investing in Primacy
The preconditions for moving to AI primacy in product design have been created by the $400 to $600 billion that has been invested into AI capabilities since 2022. This level of investment is best understood through comparison to another major step change in human technical capability: the Apollo program.
In inflation-adjusted real dollars, the Apollo program cost approximately $260 billion and took place from 1961 to 1972 (11 years). Despite being focused on space exploration, it led to an entirely new wave of technologies that reshaped life on Earth: GPS, memory foam, camera sensors, water purification systems, image processing growing into MRI, and advanced insulation materials.
If Apollo’s investment was enough to reshape the world, the investment in AI should remake it. This is because to-date, global investment is more than double what was spent on Apollo in less than one-fifth the time. Shockingly, this investment is only intensifying with projected totals reaching $1 trillion to $2.35 trillion by 2030.
Whereas the immediate goals of Apollo was extraterrestrial, AI’s explicit purpose is reshaping our world. This can be seen in some of the more exotic use-cases: discovering new materials, predicting weather, and protein folding. Beyond rebuilding how we deal with the material world it is clear a central target for the new wave of reasoning, judgment, and near-real-time execution seen in the agentic use-cases is automating human labor. Given what the core model teams have delivered in two years, anyone who thinks the existing adoption blockers signal a structural inability to automate this work is undervaluing the history of success these teams have demonstrated and the profound acceleration in capability that will result directly from this level of investment.

Why People Struggle to See this Change Coming
Humans operate on limited cognitive bandwidth and frequently compress information. We famously make decisions using 7±2 dimensions (i.e., between 5 to 9 dimensions per person/situation), and this limit defines how we reason, plan, and decide. This helps us survive the deep complexity of our world, but it also blinds us to it. We find clarity by reducing or abstracting variation, not by holding more of it.
Machines do not reason this way. Instead of juggling a handful of dimensions, they hold trillions of relationships in memory simultaneously and this allows them to spot patterns across domains that humans see as completely unrelated. For example, a person might view running a kitchen and managing a leasing office as entirely different. A machine may see them as nearly identical because both involve: prioritization under constraint, coordinating across roles, sequencing tasks, interfacing with physical environments, supporting customers, and handling exceptions.
To most people, this similarity is invisible but, with enough data and training time, it becomes obvious to machines. As such, what creates human skepticism is a failure to understand how quickly AI can identify non-obvious patterns in human-made systems, and how much faster it becomes with each new cycle. When a model stumbles, many assume the error reveals an intrinsic limitation. What is more likely true is they are seeing a training or information gap that, at this level of investment, will disappear faster than most can imagine.

Personalization Removes Customer Service Concerns
With AI-driven pattern recognition rapidly taking over service-based work, the remaining high value human differentiator cited is customer service. Critics argue that machine responses still feel synthetic, and that people prefer the warmth and nuance of a human touch. I find this argument unconvincing for two reasons.
First, when given context about the person, AI is already becoming better at understanding and managing emotion than even the most skilled therapist. Reasoning from this, I see a world where a token (similar to those around ads) is passed to the model, containing information about how a person prefers to be spoken to and other core psychological traits. With this input, the system will be able to deliver highly personalized service in a tone and structure that is emotionally precise and perfectly calibrated.

Second, people care more about speed and accuracy than about whether a human is involved. As customer service approaches real-time speed with near-perfect accuracy, the only reason people will seek human interaction is when they believe the policy itself is wrong. This is a path already closed off by many Fortune 500 companies and, for companies that want to support it, a process that can be handled by humans can be easily created then likely automated.

Designing for the Transition
Operational workflows today rely on a familiar structure: humans receive input (often from customers), follow rules, and resolve the task using custom-built tools. In this model, humans are both the primary decision-makers and the primary agents of execution. Automation exists to speed them up and reduce errors, but the system still assumes that people are the core of the process.
Designing for AI primacy starts with a different assumption. The only persistent role for a human will be initiating the action. Once triggered, the system should rely on machines to evaluate intent, interpret nuance, apply business logic, construct personalized responses (when needed), and take action.
Humans remain in the loop only when the model encounters use cases without clear policy or precedent. The solutions to these interactions (whether a fresh answer is given or a recommendation confirmed) become training signals that continuously improve the system's logic. Through this approach, a flywheel that turns every edge case into signals that clarify the system's logic and eliminate the need for human intervention is created.

Closing
This is not a debate about whether AI will replace human execution. That is already underway. The real challenge is whether product teams will start designing for what’s already possible or get left behind waiting for the obvious change to be realized. The longer teams design around humans, the more technical debt they accumulate in their code, architecture, onboarding processes, and customer expectations. My next share will provide an applied example of automation from the property management use-case to show these concepts in action.

[Modern Problem Solving @ModProblemSolv]

@GestaltedApe OOO good point. I have another bit of research i did into indoor light that i will write up and share. Thanks!

[Modern Problem Solving @ModProblemSolv]

I spend a lot of time working on calming the nervous system and learning about it. One thing i added to my approach about 2 months ago was CO2 based breath training. CO₂ is great for this because it helps gently retrain the limbic system to respond with calm instead of panic. It does this by putting pressure on from the autonomic stress of increasing CO2 levels but doing it in a controlled and safe situation. Over time, this helps reduce overactive stress responses and builds resilience, which is super cool.

Anyhow, I have found this really helpful when combined with my meditations that focus on rebuilding a sense of being embodied. There is a lot of good research here about the connection with the limbic system (brain’s emotional control center) and this taps into some research around the separate fear and anxiety system that is more ancient than the amygdala fear system (this is the part of the brain people associate with feeling negative feelings like fear). I think for people who have conditioned themselves to use executive function to suppress negative feelings, breath work is a good way to rebuild connections because the “fear/discomfort” that is being created isn’t amygdala driven and cannot be managed the same way.

Anyhow, I have found a simple 15 minute breathing session focused around CO2 retention to be a really great way to drive deep meditations but also a general sense of well being. The program can be learned in one day and is well organized around 3 body movements you do while holding breath and three styles of breath you use to start each ~5 minute session (done back to back to back). My friend Alexia has a great program for €7.77 a month that trains how to do this too (https://t.co/YHHHL7WRIs). If you want to learn more about breathing and the limbic system, you can check out a guide I pulled together from my earlier research here: https://t.co/Fv2wcy7toY

[Modern Problem Solving @ModProblemSolv]

Cool
https://t.co/T7zQE24z1H

[Modern Problem Solving @ModProblemSolv]

https://t.co/tkCZ0gJ7o1

[Modern Problem Solving @ModProblemSolv]

https://t.co/pCSpVCZQyd

[Modern Problem Solving @ModProblemSolv]

https://t.co/MhdE4vCqO7

[Modern Problem Solving @ModProblemSolv]

DIY Powerwall Builders Are Using Recycled Laptop Batteries to Power Their Homes - VICE https://t.co/YOADwzdeyf

[Modern Problem Solving @ModProblemSolv]

Details on how vaccines go from design to production. https://t.co/eS8CfVtd6r