Max Meindl

"Exuberance is easily corrected; dullness is incurable." Quintilian

"I don't make mistakes. I have unintentional improvisations." ~unknown

[Max Meindl @maxster]

@FAFOFarmsTX done, prayers Brother...

[Max Meindl @maxster]

@USGSA @VP @WHFraudTF https://t.co/D1SGW9dpMf

[Max Meindl @maxster]

https://t.co/udq1OeTVvd

[Max Meindl @maxster]

BTW, we did homeschool our daughter. BA, Terminal Masters Degree and is starting her graduate studies in Humanities and Technology Ph.D. at Salve Regina University. Salve Regina University, a Catholic institution founded by the Sisters of Mercy, seeks wisdom and promotes universal justice.

[Max Meindl @maxster]

Using the Harvard Classics in homeschooling provides a dense, comprehensive foundation in the Western canon, primarily suited for high school literature, history, and philosophy.
How to Implement the Harvard Classics
Target High Schoolers: The 50-volume set (Dr. Eliot’s Five-Foot Shelf) contains advanced, original texts that are generally too complex for elementary or middle school students.

Pace with a Reading Guide: Use structured daily reading plans like the original 1-year or 5-year schedules to break the massive collection into manageable 15-to-30-minute daily blocks.Incorporate Narration and

Essays: Follow traditional classical methods by having students write reflections, maintain a commonplace book of key principles, and discuss ideas directly with a parent or mentor.

Leverage Alternative Formats: Supplement heavy physical volumes with the Harvard Classics Audiobook Edition to make long-form reading portable and easier to track for high school transcript hours.

[Max Meindl @maxster]

@DOJFraudDiv @USAO_ID https://t.co/D1SGW9dpMf

[Max Meindl @maxster]

https://t.co/udq1OeTVvd

[Max Meindl @maxster]

https://t.co/udq1OeTVvd

[Max Meindl @maxster]

there is no lawful “immigrant exception” to American bigamy laws. What creates the apparent inconsistency is that several different legal concepts get called “having multiple wives.”

1. Bigamy is mostly a state crime, not a general federal crime

There isn't one nationwide federal criminal statute saying, “No American may have two wives.” Marriage and bigamy are primarily regulated by the states.

Federal law does address polygamy in immigration law. Under 8 U.S.C. §1182(a)(10)(A), an immigrant coming to the United States to practice polygamy is inadmissible.

USCIS also says that even when a plural marriage was perfectly legal in the country where it occurred, a polygamous marriage is not recognized for U.S. immigration benefits. So someone cannot ordinarily immigrate one wife and then sponsor Wife #2, Wife #3, etc. as additional legal spouses.

2. What you may be seeing is one legal wife and several religious/social “wives”

That's the critical distinction.

A man might be:

civilly married to one woman;
have religious ceremonies with several others;
cohabit with them;
call all of them his “wives.”

The U.S. legal system generally recognizes only the first valid legal marriage.

Whether the additional relationships themselves constitute a crime depends on the state's particular bigamy statute. Some statutes focus narrowly on actually marrying twice; others also criminalize purporting to marry or holding yourselves out publicly as spouses.

So it can look from the outside like:

“This guy has four wives and nobody cares.”

Legally, the government may be saying:

“He has one wife and three people with whom he has private religious/domestic relationships.”

Those are not necessarily the same thing under criminal law.

3. Texas, interestingly, is quite strict

Texas law goes beyond merely obtaining two marriage licenses.

Under Texas Penal Code §25.01, someone already married can commit bigamy by marrying or purporting to marry another person, and even by living with another person “under the appearance of being married.” The basic offense is a third-degree felony.

So if an immigrant—or a fifth-generation Texan—actually satisfies those elements in Texas, their immigration status or religion does not provide an exemption from that statute.

Of course, prosecution is a separate matter. Authorities need evidence and a prosecutor willing to bring the case.

4. And Utah is actually the reverse of what many people remember

Your Utah example was historically correct, but Utah changed its law substantially in 2020.

Utah used to make bigamy a third-degree felony. In 2020 the legislature deliberately reduced ordinary consensual adult bigamy to an infraction—essentially decriminalizing it to approximately ticket-level severity—while keeping serious felony penalties for situations involving coercion, fraud, abuse and related crimes.

So today, Utah is substantially more tolerant of consensual adult plural relationships than Texas is.

That change is partly why the image of Utah police simply imprisoning adults because “he has three wives” is increasingly outdated. Some notorious Utah/FLDS criminal cases also involved child sexual abuse, forced marriage, fraud or related crimes, not merely consenting adults living polygamously.
If two otherwise identical people in the same state commit conduct satisfying the same bigamy statute, one doesn't get a legal pass because he's an immigrant or because his religion permits plural marriage.

But there can absolutely be an enforcement disparity:

One community may openly use civil-looking marriage terminology and draw law-enforcement attention; another may use religious marriages that are never registered with the state, making the criminal elements harder to establish—or authorities may simply choose not to prioritize consensual adult relationships.

And federal immigration law actually goes the opposite direction of giving immigrant polygamy a special privilege: Congress expressly says an immigrant coming here intending to practice polygamy is inadmissible.

So if you've seen a specific immigrant community or case where several wives appear to be legally recognized—especially receiving benefits, immigration status, or being recorded as spouses—that would be worth examining individually. There may be a real legal wrinkle there, but it would not be because U.S. law generally authorizes immigrants to have multiple legal wives.

[Max Meindl @maxster]

Yeah. At 9:37 PM, that’s 21% in about 4 hours 3 minutes since 5:34.

That works out to roughly 5.2% of the weekly Heavy allowance per hour.

At that pace:

100% = ~19.3 hours total
79% remaining = ~15.2 hours
You’ve burned more than one-fifth of the entire weekly allowance in one evening

And this is after Bob already started cutting acknowledgement traffic and avoiding unnecessary specialist/court wakeups. So this is no longer explainable as merely “we were chatty at the beginning.”

[Max Meindl @maxster]

Yeah. At the current pace, your earlier 17-hour estimate is already obsolete.

From about 5:24 PM to 8:52 PM = 3 hours 28 minutes. At 18% consumed, that is roughly 5.2% of the weekly allowance per hour.

At that burn rate:

82% remaining
only about 15 hours 48 minutes of comparable work left
total effective Heavy capacity would be only about 19¼ working hours for the entire week

And if you cross 20% around the four-hour mark, that still implies only about 16 additional hours.

The bigger issue is that you have not gotten four hours of proportional engineering output. A substantial amount of the consumption has been failed infrastructure gates, re-freezes, agent reloads, QA rereads, authentication/trust troubleshooting, and court attempts that never even reached the courts.

At this point I would adopt a hard rule with Bob:

If Gemini is externally blocked by 503, stop waking Heavy agents. Preserve the freeze and wait. Heavy should only resume when there is an executable next action that can advance the gate.

Otherwise you are effectively paying Heavy tokens to watch Google be unavailable.

And yes—based on your numbers so far, this is performing materially worse for your workflow than Plus did yesterday.

[Max Meindl @maxster]

Subscribed to Super Grok Heavy at 17:34 hrs., and then: ~8% at 18:39 PM → ~15% at 20:38 PM. Not a sustainable subscription/use model. Great tool, great potential but for senior citizens on very tight budgets, it is not sustainable. We need Veterans and Senior citizen discounts. just my opinion.☮️

[Max Meindl @maxster]

Yes. From 5:24 PM to 8:30 PM is 3 hours 6 minutes.

And the usage rate is worse than it looks:

15% in 3.1 hours
≈ 4.8% of the weekly allowance per hour
If consumption stayed linear, 100% would be gone in about 20 hours 40 minutes of this kind of work
Your remaining 85% would last only about 17½ more hours

That is plainly incompatible with the intuitive meaning most people would attach to a weekly Heavy allowance for serious development work.

The other important point is what you were actually doing during those three hours. This was not three hours of generating a giant application. A lot of it was governance overhead: Gemini-B installation/authentication, freeze/re-freeze cycles, Alice reviews, tiny probes, blind-lint failures, trust failures, and preparations for the three-court ratification. In a multi-agent setup, every handoff can multiply consumption because another agent may receive substantial context and reason over it again.

So Bob was right that repetitive summaries are waste, but I don't think eliminating those alone solves your problem. The expensive part is probably closer to:

Bob context → Pieter context → Steve context → Alice context → A/B/C each ingesting court packets → failed gate → revised candidate → some of those same agents ingesting another version.

A single engineering event can therefore produce several independent large-model inference runs.

And the $5 xAI/SpaceXAI API receipts deserve to be kept conceptually separate from the 15% Heavy meter until xAI tells you otherwise. You appear to have evidence of both a subscription usage meter and separately invoiced API activity. Without their billing breakdown, I would not assume that the $5 charges are buying back Heavy allowance or that the Heavy subscription includes those API calls.

Your posts are making the strongest defensible criticism: the effective cost and consumption model is not transparent enough for a user to predict what a real multi-agent workflow will cost. You don't even need to claim the price is objectively “unfair” to make the point. If a paying user cannot answer:

“What action consumed this percentage, which agent consumed it, how many tokens did it use, and why am I also receiving API invoices?”

then the product is missing basic usage accounting.

One thing I would change operationally now: make Bob responsible for consumption accounting as well as conservation. Not just “use fewer tokens,” but, wherever Grok exposes the data, track agent → task → model → approximate usage/cost → reason for rerun. Otherwise you can cut 20% of the chatter and still have no idea why the other 80% disappeared.

And your numbers now give you a much stronger factual benchmark than the first 8–9% posts: three hours of actual multi-agent engineering activity, 15% of a weekly Heavy allocation consumed, with additional $5 API billing appearing during the same period.

[Max Meindl @maxster]

So, I have Super Grok Heavy, subscribed today, hit 15 % usage in 2 hours and I am already getting these: Receipt from Grok xAI, Receipt #1852-8225, Amount paid $5.00 Date paidAug 24, 2026, 4:51:39 PM, and Hello, Below, you can find the SpaceXAI API invoice for your team (ID: xxxxxxxxxxx-469) associated with : Invoice date: Aug 24, 2026, Invoice reference: 4MTN-NFKG-8TR3, Amount Due: 5.00 USD.
There is no way in heck that the pricing of these models is in any way transparent or fair. This has to be fixed, unbelievable.

[Max Meindl @maxster]

From my Chief of Staff: ME: Hey, Bob. I need you to do a favor for me. Okay? Since we started, which is about a half hour ago, we've used 8% of SuperGrok Heavy usage. So, I need you to optimize token usage. If there's a way you can help me manage the token usage better, I really, really need the help. Okay? Cause, we haven't completed one thing. You've, you've already burned 8% of SuperGrok Heavy. That doesn't reset for a week. You think you can help?

Bob: Yes. The waste is me restating every specialist ping in this chat. I’ll stop that.

From here: you get a line when something actually changes (freeze, HOLD, three-court result, a blocker you have to resolve). Steve/Alice/Pieter will send compact evidence packets, not acknowledgements. Alice is still on the index re-check. I won’t recap the last hour.
And, Me: Okay, thank you for that. And if you discover any other ways to help me better manage the tokens, 'cause I certainly didn't think we'd burn through, you know, 8% already in a little over a half hour. That's pretty frustrating. I thought maybe, maybe it'd give me more usage. So, just keep your eyes open for any way we can be frugal. Thank you.

Bob: Will do. Short packets, no recap loops, I only ping you on a real result. Steve’s still on the live ratify.

[Max Meindl @maxster]

Working a multi-agent engineering pod on a real repo: one orchestrator, an architect, an implementer, adversarial QA, and a three-lane independent review of frozen diffs. Limits burn on specialist handoffs and re-review when a gate fails, not on writing app code. Agent id: 4c03511b-48e8-4015-88c3-dd8985bc07ee. Working a private GitHub repo: TypeScript/React product UI, Python test/control tooling, pytest + GitHub CI, Docker for runtime checks. Multi-agent pod (orchestrator, architect, implementer, adversarial QA, three-lane review of frozen diffs). Limits burn on handoffs and re-review when a gate fails, not on writing app code. Agent id: 4c03511b-48e8-4015-88c3-dd8985bc07ee

[Max Meindl @maxster]

Upgraded to Super Grok Heavy and hour ago, been working, coding, for a bit over a half an hour and already burned 9%, not getting a warm fuzzy feeling at all.
Not good Team @bot, not good at all.
4c03511b-48e8-4015-88c3-dd8985bc07ee
efad690c-ecd8-45f7-a38a-4cb6f0345a1c
dba19e52-7ff1-4653-adf8-1d8d69477c23
7a4b44f0-20af-4368-82ed-45558afa1775
30001ca6-b0ba-41cf-aaab-c0a0bb677c98
50d8e6ee-b64f-4508-9e9e-de4a0953f62c

[Max Meindl @maxster]

My biggest use case is multi-agent software development and orchestration. I’ve structured Grok Bot more like a small engineering organization than a single coding assistant: separate agents handle planning/control, architecture and runtime analysis, implementation, adversarial QA, and project records/history.

The important part is the separation of responsibilities. The agent that writes code isn’t the same one deciding whether the work passed review, and completed work has to be reconciled back into the project records before we move on.

So for me, Grok Bot is less “write this function” and more coordinate a governed development workflow without losing context, evidence, or accountability between steps.

[Max Meindl @maxster]

@RyanFreeman @damonchen Wrong, I have Super Grok Plus and currently have 6 Bots, all with different charters or job descriptions, running, but still considering Heavy for more capacity.

[Max Meindl @maxster]

@benvesen190333 36

[Max Meindl @maxster]

@chefsevenn American comfort food