One founder, repeated experiments, the same ending. He builds a “democracy,” hits a crisis, reaches for powers the constitution never gave him — and each time concludes the problem was the people. The confessions are unusually honest, which is what makes the misdiagnosis legible.
Date: 2026-07-20 (reviewed) Author: Trolligarch (in-game “Calligraph”) — retrospective, ~9 years on, now holds an LLB and a BCL specialising in comparative constitutional law. That later training is why the videos can name their own constitutional violations precisely.
| Part | Video | Setting | Ends with |
|---|---|---|---|
| 1 | I founded the Habbo Roman Empire and I have a confession (55:22, 2025-10-18) — transcript · VTT | Habboon (Habbo retro), Nov 2016 | Forged democracy sustained for months |
| 2 | My school had a democracy experiment. Someone broke it (23:37, 2026-07-15) — transcript · VTT | School WhatsApp group, Feb 2017 | Collapsed in under a week |
| 3 | When teenagers run virtual democracies (1:03:18) — transcript · VTT | Cross-platform documentary (Reddit, Discord, Minecraft, Habbo, NationStates) | The general theory |
A specimen page (dated, source-anchored) for Aggregation vs. Sorting and The Gödel Governance Problem — observed institutional failure rather than theorised.
A teenager built a virtual “democracy” as a product differentiator against rival dictatorships, then hollowed it out from inside — forged every Senate vote, wrote every law himself, ran show trials on retroactive statutes, and kept the ceremonial forms intact. He accidentally re-ran the Roman Republic→Empire arc without knowing Roman history.
Setting: Habboon, an unofficial Habbo Hotel server (“retro”). Player-run “agencies” (fake CIA/MI6/police forces) competed for scarce members. Retros give the currency away free — so unlike real Habbo, money could not buy loyalty.
A specimen page (dated, source-anchored) for Aggregation vs. Sorting and The Gödel Governance Problem — an observed institutional failure rather than a theorised one.
1. Democracy as market differentiation, not conviction. Every rival agency was a dictatorship justified by first-mover status (“I built it, I paid for it”). Labour was scarce — too many agencies, too few players. Democracy was the one product competitors structurally could not copy: “you too can be the Empress of Rome.” His own retrospective admission: “When I created the Roman Empire in November 2016, I didn’t actually believe in democracy because I thought democracy was stupid.”
2. The patronage/rank-inflation arms race. Unable to pay in currency, agencies bought veteran loyalty with titles, inflating hierarchies past parody — one agency had five people ranked fifth in command, six ranked second, and seven ranked “zeroth in command” because they began counting at zero. Modern ones have a fifth-in-command outranked by 86 other people. Titles were the currency; the ladder was rigged so plebeians never reached the top.
3. The founding compromise — restricted eligibility. To poach veterans he kept universal suffrage but restricted candidacy to high-ranking senators/legionaries, and parachuted veterans into those ranks. He frames this as “constitutions are the products of compromise” — then later admits it was what he wanted anyway, and the patricians were a convenient scapegoat.
4. No codified constitution — deliberately. Partly ignorance (UK-raised; the Habbo White House had none either), but also design: “the constitutional ambiguity was kind of the whole point.” Branch powers were contradictory on the official site — an Empress “in command of everyone and everything” alongside a Senate that wrote laws and ran courts, with no document to resolve the clash.
5. Separation of powers, rediscovered from scratch. Senator Max objected that the Secretary of the Senate — an executive appointee — could not head the legislature. Calligraph had never heard the term: “Separation of powers? What’s that?” He conceded the change without understanding why it mattered.
6. Unlimited simple-majority sovereignty. The rule adopted: any change, big or small, needs a Senate bill; a simple majority passes it; no restrictions on what legislation may change — election rules, impeachment, or abolishing rights. His own later assessment: “the position of senator was absolutely busted.”
7. The Apollo show trial. A rival agency owner was tried for encouraging a Roman member to defect, and sentenced to “death” (permanent ban). Publicly it was a triumph of the rule of law. The defects, per his own account: no rules of court procedure (the judge improvised); no defence counsel; the prosecutor was also the key witness; the accused was found guilty before the trial began (the trial was the defence’s chance to disprove it); and the statute was passed retroactively and backdated to the exact date of the alleged offence.
8. The confession. “The Senate never voted for the Defection Act. The Senate never voted for the Workplace Act. The Senate never voted for a single piece of legislation. All of the Senate records were a lie.” He wrote every law himself, recorded himself as approving them, added inactive senators as fake yes-votes, and signed one bill under another person’s name. If challenged he would tell a senator they had missed the vote.
9. Legalism as cover. He enforced the unpopular laws with maximum zeal — ticketing players for being AFK at a desk — saying he had no discretion because “the law reflected the will of the Senate.” The Senate had never voted for any of it.
10. Why he subverted it — the security dilemma. Democracy created a technical vulnerability. Room rights (the ability to place furniture and kick players) were the real substrate of power: they enabled “trashings” and nukes, and had to be granted to the elected Empress or the democracy was a visible sham. He stalled, automated, and hid base ownership behind an alt — then caved. “If an agency could infiltrate the Empire with enough of its moles, it could potentially get one of their own to be elected and destroy it from within.”
11. So he fortified it — by shrinking the electorate. Rather than rig votes, he controlled who entered the powerful body. Recruits chose Legion or Senate; the Senate held all real power, so he sold the Legion with framing: a glamorous conquering war lady versus a boring do-nothing paper pusher who prosecutes people for being AFK. Then: total surveillance (a database of every citizen’s finances and affiliations, alt accounts, bugs in every room including meeting rooms), a black site (a client-crash bug to “disappear” players without a visible ban, with public denials), WMD built without the knowledge of the Secretary of Defence, the Princess, or the Empress — and a preemptive strike on the Habbo White House, a friendly ally, with no authorisation and no evidence of threat.
12. The kicker. He never ran a contested election — Vocaloid and Contrary stood unopposed. And he believed his own story: “I professed my love for democracy… In the end, I bought into my own propaganda. I believed that what I had done was democracy.”
Same founder, three months later, ~20 schoolmates. This one is the clean experiment: the constitution is short enough to quote in full, the failure is dated to the hour, and the founder is candid that he caused it. It is also where Chris’s agenda-control prediction lands exactly.
1. A real written constitution — nine clauses. One vote each; monthly presidential elections; all elections managed by the host (him); president appoints VP; only president and VP may propose laws; 50% for normal laws, 66% supermajority for constitutional laws; term limits on both offices. Plus “irreversible laws” — entrenched clauses — the operative one being that laws, decisions and actions must not affect real life.
2. He identifies the genre himself: Nomic. Peter Suber’s 1982 self-amending game, where the rules for changing the rules are themselves in play. His crucial deviation: no win condition. Nomic terminates at 100 points; his had “no theoretical end game. There was no specified way to win.”
3. Duverger at N≈20 — via pre-election merger. Four parties formed; polling immediately became a two-horse race. Consolidation was negotiated, not mechanical: “behind the scenes, there was some backroom horse trading. Players began sequestering themselves in private group chats” — and when Pat Troy launched a third left party, Ludi persuaded him to merge into the Red Tory Union. Coalition-formation reproducing the binary one level up, in a school group chat.
4. The trigger was exit, and exit was contagious. The chat flooded with campaign spam; people began leaving; WhatsApp announces every departure, so “one exit became two exits.” Visible exit produced a cascade that threatened the polity’s existence.
5. The extra-constitutional response. He issued two “executive orders” — banning images, then banning all discussion except at times he allocated — with exile as the penalty. He now enumerates his own violations: no vote (breaching §6’s majority requirement), and he was neither president nor VP (breaching §5’s monopoly on proposing laws). His retrospective best defence is a generous reading of §3 (“all elections… managed by the host”) stretched to cover campaign regulation. His honest account of what 16-year-old him would have said: “Fuck you, I’m the host.”
6. The agenda-control moment. Nikita, whose ticket 44% of respondents said they would vote for, posted one more image and had his candidacy invalidated by the host — under an order the host had no authority to issue. The frontrunner was removed by the referee. Nikita quit rather than run by proxy: “Tis better to die a lion than become a sheep.”
7. Realignment into an anti-referee bloc. The orphaned 44% had to choose between their arch-rivals and a joke party. Instead the joke party dissolved into the opposition on explicit revolutionary terms, and the axis of politics stopped being left-vs-right and became everyone-vs-the-host: “Reclaim your basic WhatsApp rights… Reclaim what is yours from the megalomaniac zealot.”
8. A player diagnosed the mechanism live, in-chat:
“This is how dictatorships form. Trolligarch wants to do one thing, the people say no. Then Trolligarch exerts complete and total authority under the guise of a constitution.”
9. The end — by admin action, not by vote. “Trolligarch removed Loody. Thank you all for participating in this failed experiment.” The substrate was WhatsApp group-admin power, which no clause of the constitution ever mentioned — exactly as room rights were in Part 1.
10. The lesson he drew. “The people clearly could not be trusted with full democracy,” he concluded eight months later — and built a second republic that “stripped all pretenses of democracy”: a general secretary appointed by the founding father; a chairman who cannot act without the general secretary’s consent; the founding father (judiciary) is him; and the first general secretary he appointed was also him. One concession: the chairman would be elected.
The documentary Parts 1–2 are personal footnotes to. It generalises across Reddit, Discord, Minecraft, Habbo and NationStates — and independently arrives at the substrate problem, then proposes the only answer that actually works.
“Why do founders have such an outsized impact on a democracy’s future? The answer is that they possess the power of a God.”
Whoever owns the Discord server / subreddit / Minecraft box can censor, ban, appoint, dismiss — and delete the whole polity. “A Discord president could insist that they were exercising their veto over and over again; that doesn’t stop the reality that a frustrated owner could delete the Discord server entirely.” This is exactly the room-rights (Part 1) and group-admin (Part 2) layer, named as the general case.
Three constitutional responses, all flawed:
| Model | Mechanism | Failure |
|---|---|---|
| Transfer ownership to the winner | server ownership follows the elected office | “basically never done” — every election becomes Russian roulette; some 15-year-old will try the coup |
| Owner outside politics (transcript renders the term as “jeel/jewel state” — uncertain, plausibly “dual state” in Fraenkel’s prerogative/normative sense) | owner handles costs + out-of-character rules, may override democracy to do so; only check is threat of revolution | revolution is a nuclear option needing broad agreement; divisive owner decisions have no appeal; requires a benevolent and competent owner — the community coined “admonocracy” for owners lacking self-restraint, and the existence of the word is the evidence |
| Judicial review of the owner | courts can check owner acts | “a poorly designed system of checks and balances can dangerously obstruct the enforcement of out-of-character rules” — see §D |
Since the substrate cannot be bound, make it replaceable: backup servers, archives of legislation/court decisions/documents duplicated by designated archivists and by ordinary users, parties, press and NGOs on infrastructure the owner doesn’t control.
“If they delete the server or commit a coup d’état, users would simply revolt against them and move to a different server like nothing ever happened.”
Exit becomes the enforcement mechanism for the constitution — not a substitute for voice, but the thing that makes voice binding.
Two elements make a virtual democracy: infrastructure (platforms) and labour (participants).
The cost of the infrastructure determines political stability. Thousands of dollars of modded Minecraft server or Habbo HQ ⇒ revolution requires recreating all of it ⇒ expensive, therefore rare. A free Discord server plus Google Docs ⇒ revolutionaries copy the documents and start over ⇒ trivial, therefore constant. “This price tag of revolution is why certain forms of sim governments are much more prone to political instability than others.”
The evidence is the contrast: Voices of Democracy ran 3.5 years to one orderly transfer of power. The Republic of Kramara (Discord) took 6 weeks to its second owner — by nuking the server — and then ran through fifteen republics, including a 1st Republic whose owner voided a congressional election and installed a member who had been present under an hour, who founded the Kramaran Nazi Party and became its first self-declared Nazi president (implying more than one) and built internment camps for citizens of a rival Discord state; a 6th Republic declared “the end of history” that balkanised five months later into two rival self-declared 7th Republics; an 11th kept alive by its owner secretly running alt accounts to simulate a functioning government across all three branches; and a 13th Republic that lasted less than two hours because the deposed Emperor still owned the server and nuked it.
Labour is the other half — and it’s skill-differentiated (poster-making vs. writing election software vs. writing court judgments). Four hypotheses for why anyone participates: pay (needs a currency), available activities (Minecraft has businesses and hitmen; Reddit has posts — hence some Discord democracies crank elections to weekly to manufacture activity), drama (VOD’s “1812” crisis tripled message volume and produced its highest-ever turnout), and social facilitation (people join because friends did). His own critique of the drama theory is the sharp one: “higher voter turnouts and higher message counts indicate more activity… they don’t indicate more productivity. There’s no guarantee that drama will bring in more judges, archivists and software developers.”
Coups need both. Habbo White House, March 2018: two admins moved hundreds of dollars of assets to an account the president couldn’t reach, and he resigned. By contrast, nuking a Discord server without popular support “is quite frankly an astonishingly terrible plan for a coup” — infrastructure seized, labour not.
Textbook democratic backsliding, executed entirely through valid procedure:
The most serious section. Allegations that a 21-year-old former president solicited explicit images from a 16-year-old (denied), amid prior cases of soliciting/disseminating images of minors, stalking and doxxing. Prosecutors faced six hurdles, all of them rights-protective by design:
Outcome: “prosecutors simply gave up trying.” A culture of “institutionalised indifference”; allegations became “mere factoids… placing the burden on victims to avoid interactions with their alleged abusers.” Evidence in a case involving a minor could be seen by ~20 teenagers, some of whom “demonstrated worrying ignorance regarding aspects like the age of consent.” The accused was finally removed only via a loophole — lobbying an allied Habbo organisation to issue an international arrest warrant so he could be banned for “complying with obligations under international law.” The 2020 SAFE Act fixed it by removing the process: zero-tolerance “one-punishment crimes” with mandatory permanent bans, discreet evidence handling, and the ethics board replaced by direct admin banning.
“This example is great because of what it illustrates… though there were pressures on the 3rd part[y] to join — Duverger. And it was fun to see the group ‘rebelling against a tyrant!’”
Part 2 contains two third-party absorptions in a single week-long game, both by negotiated merger rather than by voters abandoning a hopeless candidate:
This matters for Aggregation vs. Sorting §2: it is Duverger’s psychological effect operating on elites rather than voters, and at N≈20 — the pressure landed on party leaders deciding whether to exist, not on citizens deciding whether to waste a vote. It also confirms Chris’s earlier point that coalition-formation doesn’t escape the binary, it relocates it: the merger happened before the election, not after.
And the second absorption is the more interesting one, because it shows the axis rotating in real time — Duracell didn’t join the Red Tory Union on policy (it was founded to oppose their drug legalisation plank), it joined an anti-referee bloc. That is the realignment/change-of-basis mechanism from the thesis’s §6, observed at small N and over about 48 hours.
“The highest critics come from the libertarian side — Caplan, The Myth of the Rational Voter, and Hoppe, Democracy: The God That Failed. But it shows the Gödel strange loop empirically. Granted these were done in virtual environments by children who were motivated to participate, but I think it shows how difficult it is to create a constitution and keep it intact. The founding fathers have to be given props for creating a system that has lasted 250+ years. These republics died in weeks/months!”
(Bibliographic note: Bryan Caplan, 2007. The “Caplan” already in this vault — the 1984 Toronto debate — is Gerald Caplan, Canadian NDP, an unrelated person. No cross-link. Hoppe is already present via SAVE Act and Suffrage.)
The corpus splits the two critics — and it goes badly for Caplan’s mechanism.
On the founders — the props are earned, and Trolligarch’s own theory says why. Every virtual constitution in the series failed at an unnamed substrate: room rights, group-admin powers, server ownership. The framers named theirs — force — and bound it in the text. The sharpest instance is Article I, §8, cl. 12: “To raise and support Armies, but no Appropriation of Money to that Use shall be for a longer Term than two Years” — a structural guarantee that no single Congress can permanently fund a standing army, so the instrument of force dissolves without recurring consent. That is precisely the clause no teenage constitution ever wrote. (Note the erosion: a 1904 Solicitor General opinion read “raise and support” narrowly enough to put equipping the army outside the two-year limit.)
Two complications worth holding against the 250-year claim:
Correction to the above: the exit point is not a special fact about the US, and the vault already holds it in general form. Government Formation §”The ratchet”: organisations have asymmetric incentives around membership, so they systematically make exit costly — “nations make secession illegal,” unions run closed shops, HOAs attach to the deed, the EU made Brexit maximally painful to deter others. Chris notes the vault already treats the Civil War as the end of free association.
So the convergence is the finding: the ratchet says organisations deliberately raise exit costs to escape market discipline; Trolligarch’s price of revolution shows that where exit is naturally free, no institution can stabilise at all. Same variable, opposite directions — and together they suggest exit cost isn’t a defect of institutions, it’s a precondition for them.
Chris’s sharpest observation on Part 2:
“Even in the 2nd example, he tried to ‘fork’ by decree, but was labeled a tyrant and eventually got the entire project shut down. Unlike the US, he lost that war.”
This is force-doctrine applied to constitutional founding. Both Trolligarch and the Philadelphia Convention acted beyond their mandate by unilateral decree, and both were illegal ex ante — the Convention was authorised only to revise the Articles, and Article VII’s nine-state rule contradicted the unanimity requirement it was replacing. What separates “founding” from “coup” is not the legality of the act, which was equally defective in both, but who won. Trolligarch was labelled a tyrant, lost the room, and dissolved the polity; the framers were ratified and became founders. See Force Doctrine — noting its own guard: this is fit → durable might, a descriptive claim about how the label gets assigned in retrospect, not might-makes-right. Chris’s corollary: the founding was never fully settled either — “even today we are still arguing federalist/anti-federalist talking points.” Settled enough to function is not the same as settled.
What the corpus says “a solid constitution” is — and isn’t. Chris’s conclusion (“I am still in favour of representative governments, but all of this illustrates the need for a solid constitution”) survives, but the series redirects what solidity consists of, because text quality is close to irrelevant in the evidence:
What did track durability:
| Variable | Evidence |
|---|---|
| Naming and binding the substrate | every virtual failure hit an unnamed admin layer; the framers bound force (Art. I §8 cl. 12) |
| Exit cost / price of revolution | costly Habbo & Minecraft polities endured; free Discord produced 15 Kramaran republics |
| Institutional memory | Part 3’s own thesis — every polity re-derives every lesson from zero |
So “solid” does not mean better clauses. It means: identify what actually holds power and bind it; make the polity expensive to leave; and carry the memory forward. That is an actionable redirection of the conclusion rather than a refutation of it — and note the third item is what this vault is for.