How D-Central sources, qualifies, and verifies mining-hardware reliability data

We maintain a growing, openly-documented dataset of ASIC miner failure modes, repair costs, and used-market prices. Here is exactly how it works — and where its limits are.

Where the data comes from. This dataset is aggregated from public sources: community forums (BitcoinTalk), open-source project issue trackers (GitHub), independent teardown and repair channels, public repair-shop service menus, manufacturer support documentation, and secondary-market marketplaces. It is built on the shoulders of the miners, repair technicians, and open-source builders who shared their experience publicly. We are grateful to them.

What it is NOT. It is not a tally of D-Central’s internal repair tickets. We do not publish proprietary in-house failure counts, repair volumes, or MTBF figures, because we do not measure failure rates in a way that could honestly produce them. When we say a fault is “frequently reported,” that means it is frequently described in public sources — it is a measure of how often people talk about it, not a measured probability that your unit will fail.

How every claim is qualified. Each failure mode carries an honest frequency qualifier — frequently reported, occasionally reported, or single anecdotal report — plus at least one source. We distinguish verifiable fact (a chip part number, a board’s chip count, a rated voltage) from reported experience (which components people say fail most). When a third party publishes a number — for example, a repair guide’s “30% failure rate” estimate — we quote it with attribution and treat it as that source’s claim, never as established fact. We do not invent percentages, counts, or rates.

How specs are verified. Every chip and board specification is cross-checked against our internal technical reference library before publication. Where sources conflict, we resolve in favour of the most detailed verified record and document the resolution. Where a layout is genuinely uncertain, we say so and mark it for verification rather than guessing.

How prices are handled. Used and refurbished prices are captured with their original currency, condition (new / refurb / used — never blurred), and capture date. Any Canadian dollar figure we show alongside a foreign-currency listing is an FX-converted estimate only: it excludes duty, GST/HST/PST, customs brokerage, and shipping, and it is not a D-Central price quote. We report ranges where the market is genuinely spread, and we say “new prices only” when no honest used price exists rather than inventing one.

What we will and won’t claim.

• We will cite sources, date our data, qualify frequency honestly, separate fact from anecdote, and correct ourselves in public when we get something wrong.
• We won’t imply proprietary internal numbers, invent failure rates or MTBF, present FX estimates as quotes, publish security or reverse-engineering findings targeting any specific product, or position ourselves above the people whose work we build on.

Why this exists. Good repair-vs-replace decisions need honest data, and the Canadian market in particular lacks a price signal that accounts for landed cost. We are building one in the open. If you have first-hand field data, you can contribute it — see our Field Reports program. The more honestly-sourced reports we gather, the better this gets for everyone.

Last updated: 2026-06-09. Data is aggregated from public sources and refreshed on an ongoing basis. Corrections welcome.

---

This methodology governs every dataset in the D-Central Open Mining Data hub. All datasets are released under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence.