At 7:00 p.m. KST on February 6, Bithumb attempted what should have been a routine promotional payout: a total of ₩620,000 distributed to event participants. Within minutes, customers saw something else—Bitcoin balances so large they looked like a screenshot hoax.

They weren’t.

A unit-entry mistake credited users with 620,000 BTC instead of ₩620,000 total, a misfire that briefly produced tradable balances inside Bithumb’s internal ledger. Regulators later said 86 users sold about 1,788 BTC before the exchange completed restrictions. That detail is the real story—not because 1,788 BTC is the full scale of the incident, but because it shows how quickly a bookkeeping error can become market action when a platform’s database layer functions like settlement.

This wasn’t simply a clerical error. It was a live demonstration of how a centralized exchange can temporarily manufacture tradable assets through internal bookkeeping—what Korean coverage quickly framed as “ghost coins”—and how quickly that internal accounting layer can spill into real price formation.

Timeline: the 40 minutes that mattered (KST)

Bithumb has now provided the clearest sequence in official disclosures, corroborated by subsequent reporting:

• 19:00 — Promotional rewards credited to 695 users.
• ~19:20 — Bithumb detects the abnormal crediting.
• ~19:35 — Trading/withdrawal restrictions begin for affected accounts.
• ~19:40 — Restrictions completed.

Before the freeze: regulators later said 86 users sold ~1,788 BTC in the brief window.

After the freeze: Bithumb says it reversed ~99.7% of erroneous credits and—crucially—covered the already-sold 1,788 BTC using company-held assets to restore “balance consistency.”

What failed: not “human error,” but control architecture

The operator error is straightforward: “620,000” was entered as BTC units rather than KRW. The failure is what happened next—the system allowed the resulting balances to become actionable.

A payout pipeline capable of touching market-tradable balances typically has “fail closed” controls:

• Unit validation (KRW vs BTC cannot be ambiguous)
• Hard caps (per user, per campaign, per minute)
• Maker–checker approvals (two authorized individuals for high-risk actions)
• Extreme-value alerts that automatically block execution
• Reserve-/risk-aware constraints that prevent crediting beyond inventory or limits

Yet the Bithumb incident reached customer balances and remained tradable long enough for executed sells. That’s why Korean coverage quickly moved past “fat-finger” language: the scandal wasn’t that someone mistyped—it’s that the platform’s design made the mistype market-relevant.

“Ghost coins” aren’t a metaphor. They’re a settlement problem.

Centralized exchanges operate two realities:

• On-chain reality: coins exist on the blockchain and move with deposits/withdrawals.
• Off-chain reality: balances live in the exchange’s internal ledger, where most trading occurs as claims swap hands.

This structure is normal. But it creates a specific risk: if the ledger is wrong, the market can still trade on that wrongness.

The Bithumb episode appears to be primarily an internal-ledger incident, but it still produced real effects because ledger credits could be sold into real order books. Once that happens, an exchange faces three ugly choices:

1. unwind trades (complex, legally risky),
2. compensate counterparties (costly, politically visible), or
3. absorb losses to restore balance consistency.

Bithumb’s own post-incident stance aligns with option (3): it says the 1,788 BTC sold during the window was offset using company assets.

That is why the headline number (620,000 BTC) is almost a distraction. The consequential number is 1,788 BTC—the amount that allegedly crossed from ghost credit into executed market trades before controls fully shut the pathway.

What’s new this week: compensation terms, regulator escalation, and Parliament’s focus

Compensation (now defined, not just promised)

Reporting converges on a package built around three pillars:

• ₩20,000 payment to users who were logged in during the incident period
• For users who sold at abnormally low prices during the incident window: 100% of the price difference reimbursed + an additional 10% consolation amount
• 0% trading fees for 7 days (exchange-announced fee relief window)

The watchdog response is no longer observational

South Korea’s Financial Supervisory Service (FSS) has escalated beyond routine monitoring into a targeted probe posture, with public remarks framing the incident as exposing “structural weaknesses” in virtual-asset systems that could disrupt market order.

Parliament’s scrutiny: “why wasn’t this impossible?”

The National Assembly hearing focus, as described in coverage, isn’t “who typed the value,” but why guardrails didn’t block a KRW-vs-BTC unit error from producing market-active balances—a classic Korean post-incident accountability pattern: controls are judged by whether the failure mode was structurally preventable, not merely operationally regrettable.

Why this hit a Korean nerve: trust is cultural infrastructure

Korea’s financial scandals don’t fade as trivia; they become institutional memory. Retail investors expect speed and convenience from digital platforms, but increasingly expect bank-grade safety as crypto normalizes. Exchanges, meanwhile, still operate with vertically integrated power—broker, custodian, and internal settlement system in one. Traditional securities markets separate those roles to reduce single-point failures; crypto platforms often do not.

Bithumb’s glitch forced that structural difference into public view: a database layer that can briefly behave like a settlement layer is not just a tech quirk—it is market infrastructure.

The unresolved questions that will determine the verdict

1. Audit-grade transparency: A single reconciled, third-party-style account separating internal reversals, internal trades, and any asset movements across other rails (cash withdrawals, cross-asset conversions) would materially improve trust; without it, the public must triangulate from press accounts and exchange statements.
2. Trade integrity and harm accounting: Who was harmed beyond the credited recipients—counterparties filled at distorted prices, stop orders triggered, forced liquidations during the dislocation? The final compensation scope will signal how broadly harm is acknowledged.
3. Regulatory remediation timeline: Korea’s credibility test is not whether it investigates—Korea investigates—it’s whether a clear control baseline (caps, maker–checker, unit validation, automated sanity checks, tighter reconciliations) becomes enforceable, with deadlines and consequences.

What this incident is really a warning about

If Korea’s crypto market wants to be treated like financial infrastructure, it cannot run on “best effort” controls. Bithumb’s error demonstrated that the most dangerous moment isn’t the typo—it’s the interval when the system allows the typo to become market action.

In that sense, February 6 wasn’t an outlier. It was a stress test—and the market saw the failure mode in real time.

“Bitcoin crypto currency 8K wallpaper” by Satheesh Sankaran, licensed under CC BY-SA 2.0 (via Wikimedia Commons).