Two market crises landed almost back-to-back in Oct 2025: a violent crypto flash crash and a major AWS outage. One was sparked by geopolitics, the other by technical failure. Both exposed the same thing: hidden chokepoints, fragile trust, and the difference between what you think you own and what you actually control.

In crypto, if you do not hold the keys, you do not own the asset. In the cloud, if your stack lives on one provider, your uptime is similarly rented, not owned. In British Columbia, the recent Richmond city land title ruling raised a more unsettling question: if courts can recognize a superior title beneath your deed, how solid is your ownership at all? These events together highlight a single story about blockchain resilience, centralization, and skin in the game.

On Friday, October 10, a tariff threat landed after traditional markets had closed, and crypto, which never closes and has no circuit breakers, absorbed the full shock. Bitcoin slid quickly, altcoins faster, leverage unwound, and ~$19 billion in positions vaporized in minutes. Nerves were raw, liquidity was thin, and automated liquidations amplified the drop. The magnitude was brutal, but what mattered most was what did, and did not break.

On-chain performance. Core decentralized networks kept producing blocks. The standout was Solana (peak ≈100k TPS at sub-cent tx fees). Long criticized for congestion in past cycles, it handled extraordinary load at high throughput while maintaining usability. The contrast to prior stress periods shows that recent client and network upgrades changed real-world behavior under pressure. Ethereum mainnet also remained live and secure as designed, but its limited base-layer throughput fees spiked (peak L1: 15-18 tps at avg tx fee ≈$2.62/tx; Base L2 generally stayed stable in tx fees/usage). In the chaos of bridging and swapping, some activity shifted elsewhere. Trading migrated to centralized venues or to higher-throughput chains, simply because blockspace on Ethereum got very expensive during high traffic congestion periods, like a cascading liquidation event.

Centralized platforms. A string of large exchanges, private messaging apps, and AI services stumbled or paused functionality during the chaos. Users who assumed they could act at the worst hour learned a familiar lesson: centralized venues have rate limits, safety switches, and scaling ceilings that can all trip under correlated stress. Crypto’s “all gas, no brakes” ethos meant markets kept clearing somewhere, but not everyone could access platforms like Coinbase when they needed to most. When the dust settled and rhetoric cooled, prices recovered partway, but damage was done. Decentralized L1s kept going. Some centralized on-ramps did not.

A week later, a multi-region AWS incident knocked popular consumer apps and financial platforms offline. It also stress-tested Web3’s operational decentralization. The base-layer chains, which have globally distributed nodes, kept finalizing blocks. That is the point of decentralized consensus. But the outage exposed a lot of single-vendor and single-region dependencies across the rest of the stack.

L1 chains. Bitcoin and Ethereum continued to function. Solana’s validator set, diverse across operators and facilities, also kept producing blocks. When enough nodes and stake are geographically and operationally diversified, one cloud’s bad day becomes a minor annoyance instead of a systemic failure.

L2 rollups. Here, the differences were more obvious. Base, which ran its sequencer on AWS, saw an immediate throughput drop when US-East-1 had problems. Arbitrum and Optimism, which had multi-region or multi-cloud failover, sailed through. The message was simple: decentralize the sequencer path, or at least engineer redundancy across providers and regions. A rollup that relies on a single machine in a single cloud is a single-point-of-failure system, even if settlement is built on top of decentralized Ethereum’s Layer 1.

Wallets, RPC, and the interface layer. Many users saw empty balances when cloud-hosted RPC gateways hiccuped. The ledger was fine, but the window into it was not. This is the chokepoint problem in miniature. We decentralize consensus, then re-centralize the access layer in DNS, app stores, CDNs, and cloud APIs. The outage made that visible to everyone in one morning, and panicked a lot of users who thought their funds essentially disappeared. The lesson is not to abandon convenience, but to add redundancy and keep a clean path that does not rely on one company’s dashboard staying green.

Sufficient decentralization. The goal is not maximalism at any cost. It is to arrange systems so that no single company or government can stop the core function. On the outage day, L1 block production passed that test. Some rollups and most centralized platforms did not. The remediation path is clear: diversify sequencers, add escape hatches, deploy across multiple regions and providers, and invest in decentralized physical infrastructure where it matters most.

At almost the same time, a BC Supreme Court decision declared that Indigenous title exists over parts of Richmond, with the court recognizing the Cowichan Tribes’ title under long-standing legal principles that require an exceptionally high threshold to extinguish Aboriginal title. The judgment did not directly sue private owners, but it did declare certain Crown and municipal interests defective and invalid, which places a fog over the predictability of ownership where superior Aboriginal title is recognized (95% of the province of BC is unceded!). For homeowners, businesses, lenders, and municipalities, property ownership in a G7 nation like Canada suddenly became a risk. Recognizing Aboriginal title and protecting everyday homeowners are both necessary. The burden to reconcile sits with the state, and this ruling failed miserably in execution.

What changed. British Columbia’s Torrens system promises indefeasible title. The court concluded that statutory frameworks cannot be read to extinguish constitutionally protected rights unless Parliament clearly intended it. Combined with UNDRIP commitments, the result is a recognition that registered title can coexist with, or be subject to, Aboriginal title that was never surrendered. That is a profound shift in how urban land security is perceived. It is also the first time a dense, developed urban area has seen a court recognize present title in favor of an Indigenous claim that predates a century of subsequent transfers.

Why this rattles trust. The value of property rights is predictability. If a good-faith buyer can wake up to a superior title claim emerging from the 1800s, the collateral chain behind mortgages, title insurance, and municipal finance becomes uncertain. Banks freeze. Insurers balk. Homeowners cannot sell or refinance. Even if eventual coexistence is negotiated, years of limbo become a tax on confidence, liquidity, and civic cohesion. Either the province funds compensation and revenue sharing, or markets will price uncertainty into every future mortgage.

Who pays and how. The state created the problem through historic failure to acquire land by treaty and to compensate original owners. If reconciliation is real, the state must fund solutions that do not expropriate unsuspecting citizens through legal ambiguity. There are templates, such as affirming title while acknowledging existing homes and critical infrastructure, combined with revenue sharing or compensation funds. But speed matters. A decade of litigation uncertainty would turn a legal principle into a slow-motion social stress test.

Second and third-order effects. Capital is skittish. Entrepreneurs relocate when the rule of law looks conditional. Ordinary families lose trust when their deed is subject to change after generations. Political rhetoric hardens on all sides. None of that helps actual reconciliation, and it corrodes the cultural covenant that underwrites investment and belief in individual property rights, in the first place.

Against this legal backdrop, the provincial government moved to permanently ban new crypto mining operations to the public grid, and to cap power for AI and data centers through an allocation process. The official framing prioritized industries that score higher on immediate local jobs per megawatt, such as LNG and traditional mining, over “emerging industries.”

Mining and flexible load. BC has abundant hydro, which means clean electricity. Banning rather than integrating miners discards a flexible demand resource that other jurisdictions use to balance the grid. Texas, for example, has paid miners to curtail in peak demand, smoothing loads and saving consumers. If you think Bitcoin mining is frivolous, you will celebrate the ban. If you think blockchains are becoming part of financial and compute infrastructure, you will see an opportunity squandered. My belief is that prohibition tends to export activity and in this case, important technological progress.

AI caps send a signal. AI clusters often employ fewer workers per kilowatt initially, but they can catalyze entire ecosystems. Talent follows compute. If your policy says the future must ask permission while the past gets priority, you do not become a hub for the next S-curve. You become a customer of those who did.

Short term optics, long term cost. Measuring projects only by immediate headcount misses indirect value creation, knowledge spillovers, and the compounding effects of anchoring high-skill industries. It also clashes with the theme of this post. If you want resilience, you want to own a piece of the future’s infrastructure rather than rent access to it elsewhere. The Government of BC is intentionally pushing away crypto and AI-related innovation from the province and the country — a completely backwards and asinine position for a country keen on revitalizing its tech sector.

Put the three stories together and some lessons emerge.

1. In crypto, ownership is operational. If you hold keys on a resilient L1 and have multiple execution paths, you can act when centralized exchanges or RPC gateways freeze. If your trading stack depends on one company’s servers, your market access is at their mercy. The crash and the outage week proved the point twice. Ethereum L1 still needs to scale solutions to minimize gas fee spikes. Base L2 and other RPC gateways need to decentralize their cloud hosting to prevent centralized outages from stuttering operations.

2. In cloud, resilience is a design choice. Multi-cloud, multi-region, and decentralization of critical components are not luxuries, but lifelines. Rollups that diversified sequencers suffered little. Wallets that had redundant providers kept user trust. Those that did not, showed phantom zero balances and added to user panic as the markets drew blood.

3. In land and law, predictability is the product. A callback to a previous post, “insurance is everything”. Torrens title is a promise that what you buy is what you own. When courts attempt to correct historic injustices, the state needs to pair recognition with clarity and compensation fast, or it undermines the very trust it is supposed to restore. If titles are contingent, the state effectively fails at honoring its insurance of property ownership. Investors and families will act accordingly. This means lower risk appetite, a complete shattering in the belief of property rights and ownership, and significantly more brain drain of talent and investment leaving the country.

Across domains, the common failure mode is reliance on a central actor without redundancy, and without clear recourse. The common success pattern is spreading authority, aligning incentives, and taking real ownership of the parts that matter most.

Skin in the game means you carry the blast radius of your own choices. When you rent the critical parts, someone else holds the off switch. When you own them, you absorb the risk and keep the agency. Keys, compute, title. That is the triad.

• Keys. If your assets sit behind someone else’s login, you own a claim, not the coin. Hold the keys. Maintain redundant routes to settle, trade, and view balances that do not depend on a single RPC, DNS entry, or cloud region.

• Compute. If your chain, rollup, or app is pinned to one cloud or one region, you own throughput until that vendor fails. Run sequencers and services across providers and regions: have backups and redirections.

• Title. If the legal system can recognize a superior claim that pre-dates your deed, then ownership is only as good as the state’s enforcement. The state must pair recognition with absolute certainty, or risk losing credibility.

Between both October crashes, Coinbase & Base’s lessons are straightforward: continue on path to decentralization and 100% uptime. British Columbia’s lesson is harder: reconcile historic rights without putting the whole country into legal fog. Bans on new compute and infrastructure only further outsource the future and tell innovators to go elsewhere.

Skin in the game is alignment. Builders hold keys and diversify infra because they cannot afford downtime. Courts and governments work towards reliability and clarity because they cannot afford chaos. If any one of these actors shirks the cost of true ownership, or oversteps their boundaries, someone downstream pays for it with lost access, lost uptime, and lost trust.

In the Network State era, there’s a maturing focus on people – how to attract them, organize them, inspire them, and yes, govern them. Parallel Citizen is a media blog dedicated to providing you the latest updates in promising network enclaves, network archipelagos, and network states. Subscribe for free to receive new posts and support me.

Subscribe

• My guide: Decentralized Social Media 101: A Creator Flywheel. I’ve made a step-by-step guide with everything you need to set up and interoperate Farcaster, Paragraph, and Zora in one sitting.

• Follow on Zora. Own my creator coin, get this product for free (see product page for details)

• You can also collect this post as an NFT token here: $LUDDITES

• Trezor - Open-source hardware wallets for sovereign custody

💥 Oct 10–11: Tariff shock triggers a record $19B liquidation cascade. L1s kept ticking, fees spiked where blockspace was scarce.
☁️ Oct 20: AWS US-EAST-1 hiccups. Many apps blinked out. Base showed high RPC latency.
⚡ Policy miss: Banning flexible load and capping AI sends talent and compute away. Own a slice of the future, don’t rent it.
🧱 Through-line: Reduce chokepoints, raise resilience, take real ownership.
✅ Call to action: Keys, compute, title. Diversify, failover, and demand clarity.