Bitcoin soft fork upgrades: Taproot, BIPs and what comes next
Bitcoin soft fork upgrades explained: from BIP to activation, what Taproot changed and why it took 8.2% of new outputs on Nov 12, 2026. BIP 54 is next.

Contents
- What is a soft fork, and how is it different from a hard fork?
- How an idea becomes a rule: the BIP process
- How past upgrades were activated
- What Taproot changed
- What Taproot adoption looked like on November 12, 2026
- What's next: BIP 54, CTV, OP_CAT and quantum resistance
- Why upgrades are so slow: the lesson of BIP 110
- Common mistakes
- FAQ
Bitcoin soft fork upgrades are how the protocol changes: a soft fork tightens the rules in a backward-compatible way, so old nodes keep working while upgraded nodes start enforcing extra conditions. Every change follows the same route: an idea, public review, a BIP document, code, miner signaling and activation at a block height known in advance. The most recent one to go through was Taproot, which activated at block 709,632 on November 14, 2021. Here is how that route works, what Taproot changed and which proposals are waiting in line.
What is a soft fork, and how is it different from a hard fork?
Bitcoin has no company in charge and no "update everyone" button. The rules of the network are whatever each full node checks: block size, signature format, the block reward. There are two ways to change them.
A soft fork narrows the rules. Everything valid after the upgrade was also valid before it, so a node running old software still accepts the new blocks and stays on the same chain. It simply doesn't understand the new features and doesn't check them.
A hard fork loosens or replaces rules. Old nodes reject the new blocks, and if part of the network doesn't upgrade, the chain splits into two coins. That is how Bitcoin Cash appeared in 2017.
| Soft fork | Hard fork | |
|---|---|---|
| What happens to the rules | They get stricter | They change or loosen |
| Old nodes | Stay on the network | Drop off unless they upgrade |
| Risk of a split into two coins | Low with miner support | High without full agreement |
| Bitcoin examples | P2SH, SegWit, Taproot | Bitcoin Cash (a separate network) |
Bitcoin developers prefer soft forks because nobody is forced to upgrade by a deadline. For the basics of how nodes and miners agree on one history, see What is blockchain.
How an idea becomes a rule: the BIP process
A BIP (Bitcoin Improvement Proposal) is a document describing a change. BIPs live in a public GitHub repository, and the process itself is set out in BIP 3, which replaced the older BIP 2. The steps look like this.
- Idea and discussion. The author posts the proposal to the developer mailing list and collects objections.
- Draft. Editors assign a number and the document gets Draft status. The number is an editorial step, not an endorsement: BIP 2 said plainly that people may adopt a BIP or not.
- Finished specification. Once there is a reference implementation and test vectors, the author moves the document to Complete.
- Code in a client. The change goes through review and lands in Bitcoin Core or another client, at first without being switched on for mainnet.
- Activation. Parameters are chosen: when miners start signaling, what threshold is needed and when the rules take effect.
- Deployed. After activation the document is marked Deployed. If the proposal goes nowhere, it is marked Closed.
None of these steps makes a change mandatory. The rules apply only on nodes whose owners chose to install the new software.
How past upgrades were activated
The activation method changed along with the network, usually after something went wrong. Bitcoin Optech keeps a detailed history.
| Upgrade | What it added | How it was activated | When it went live |
|---|---|---|---|
| P2SH (BIP 16) | Script addresses, multisig | Hardcoded date plus a miner poll | April 1, 2012 |
| BIP 66 | Strict signature encoding | 950 of 1,000 blocks (95%) | July 4, 2015, block 363,725 |
| SegWit (BIP 141) | Signatures moved out of the transaction body, groundwork for Lightning | BIP 9: 95% of blocks in a 2,016-block period | August 24, 2017, block 481,824 |
| Taproot (BIP 340–342) | Schnorr signatures, hidden scripts | Speedy Trial: 1,815 of 2,016 blocks (90%) | November 14, 2021, block 709,632 |
SegWit was the hardest. According to Optech, miner support stalled well below 95%, and some users launched BIP 148: nodes that would reject non-signaling blocks from a set date. BIP 91 followed with an 80% threshold, it activated, and SegWit locked in. That dispute led to BIP 8, which includes a mode where miners cannot block a change simply by doing nothing.
Taproot used a shorter approach called Speedy Trial. BIP 341 records the parameters: signaling from April 24 to August 11, 2021, a 90% threshold instead of the usual 95%, and a minimum activation height of 709,632. Miners hit the threshold early, and the rules took effect at block 709,632 on November 14, 2021.
What Taproot changed
Taproot is three documents: BIP 340 (Schnorr signatures), BIP 341 (a new output type) and BIP 342 (script rules inside it). BIP 341 was written by Pieter Wuille, Jonas Nick and Anthony Towns.
The core idea is that a coin can be spent in two ways.
- Key path. A single Schnorr signature. Most coins are spent this way, and an observer cannot tell whether the output had other conditions.
- Script path. The owner reveals one of the pre-committed conditions and a proof that it was part of the tree. The other branches stay hidden.
In practice this means:
- a multisig can be collapsed into one key, so on-chain it looks like an ordinary payment;
- complex conditions take less block space, because only the branch that was used gets published;
- the format reserves leaf versions, so later soft forks can add new opcodes without a new address type.
Taproot addresses start with bc1p. How they differ from bc1q, 3… and 1… is covered in Bitcoin address types. Lightning Network channels benefit too: opening a channel and closing it cooperatively can look like a regular payment.
What Taproot adoption looked like on November 12, 2026
We took 12 consecutive blocks, heights 975,519 to 975,530, mined between 04:20 and 07:50 UTC on November 12, 2026, and sorted every output by type. Raw blocks came from the mempool.space API; the count is ours.
| Output type | Outputs | Share by count | Share by BTC value |
|---|---|---|---|
P2WPKH (SegWit, bc1q) |
71,695 | 58.0% | 41.3% |
| OP_RETURN (data) | 21,935 | 17.7% | 0.0% |
P2PKH (legacy, 1…) |
10,765 | 8.7% | 13.4% |
P2TR (Taproot, bc1p) |
10,127 | 8.2% | 20.6% |
P2SH (3…) |
6,612 | 5.3% | 1.4% |
| P2WSH (SegWit scripts) | 2,458 | 2.0% | 23.2% |
Those blocks held 47,318 transactions and 123,637 outputs. At least one Taproot output appeared in 4,401 transactions, or 9.3%.
What that tells us: five years after activation, roughly one in twelve new outputs goes to Taproot, though its share of value is higher at about a fifth. The network's main format is still 2017-era SegWit. Twelve blocks cover three and a half hours and the share swings from block to block, so treat this as a snapshot, not a yearly average.
For context at the same moment: a fast confirmation cost 2 sat/vB, and BTC traded near $73,643 as of 3:00 AM ET. Live numbers are on the dashboard and the Bitcoin page; how fees are calculated is explained in the mempool fees guide.
What's next: BIP 54, CTV, OP_CAT and quantum resistance
No new consensus rules have been switched on for mainnet since Taproot. There is no shortage of proposals, though, and they sit at different stages (BIP repository data as of November 12, 2026).
| Proposal | What it does | Status |
|---|---|---|
| BIP 54, Consensus Cleanup | Closes four long-standing weaknesses | Complete, no activation parameters |
| BIP 119, OP_CHECKTEMPLATEVERIFY | Covenants: a coin can only be spent by a pre-defined transaction | Draft |
| BIP 347, OP_CAT | Joins two stack items in Taproot scripts | Complete, not activated |
| BIP 360, P2MR | A new output type without the quantum-vulnerable key path spend | Draft |
| BIP 110, Reduced Data | Temporary limits on data in transactions | Closed |
BIP 54 is the dullest of the group, which makes it the likeliest candidate. Authors Antoine Poinsot and Matt Corallo add nothing new and fix four things:
- the timewarp attack, which the document says could let miners drive difficulty to its minimum within 38 days;
- blocks that take too long to validate, by capping signature operations at 2,500 per transaction;
- transactions of exactly 64 bytes, which allow fake inclusion proofs against light wallets;
- duplicate coinbase transactions.
Version 1.0.0 of the document was published on May 22, 2026. Bitcoin Core has an open pull request #35793, filed July 24, 2026, that adds the rules without mainnet activation. There is no activation date.
BIP 360 is often described as "quantum protection", but the document itself is careful: it only guards against attacks on keys that have been exposed for a long time. Post-quantum signatures are left to separate future proposals.
Why upgrades are so slow: the lesson of BIP 110
In the summer of 2026 the network showed what happens when activation is attempted without broad agreement. BIP 110 proposed limiting data in transactions for about a year. Its threshold was lowered to 55% (1,109 of 2,016 blocks), and from block 961,632 nodes running the code began rejecting blocks that did not signal.
What followed, per KuCoin's summary of August 10, 2026:
- in the last period before mandatory signaling, 51 of 2,016 blocks signaled support, about 2.53%;
- the split-off branch produced two blocks, 961,632 and 961,633, then stopped;
- by then the main chain had reached block 961,744.
On August 9, 2026 the document was moved to Closed, noting a chain split with stalled mining.
The takeaway is simple. A soft fork is safe only while an overwhelming majority of hashrate and economically important nodes (exchanges, wallets, payment processors) stand behind it. That is why thresholds are high, debates run for years, and proposals with open disputes sit in Draft for a long time.
Common mistakes
- "Miners vote on upgrades." A signal in a block says "I'm ready", it isn't a vote. Nodes enforce the rules, and a miner who breaks them just loses the reward.
- "After a soft fork you have to change your wallet or coins." Old addresses keep working. You can move to the new format when your wallet supports it, or never.
- "A BIP number means the change was accepted." Drafts get numbers too. Look at the status: Draft, Complete, Deployed or Closed.
- "Taproot made bitcoin anonymous." It hides unused spending conditions. Amounts and addresses remain public.
- "An upgrade will lift the price." Nothing ties activation to the price. Taproot went live in November 2021, near the cycle top, and the market fell through the following year.
FAQ
Do bitcoin holders need to do anything during a soft fork? Usually not. Coins on old addresses remain spendable. If you run a full node, upgrade the software so it enforces the new rules itself.
Can a soft fork change the 21 million cap? No. Raising issuance loosens the rules, which makes it a hard fork. Nodes that don't upgrade would reject those blocks.
When is the next Bitcoin upgrade? No date is set. As of November 12, 2026, BIP 54 has a finished specification and an open code pull request in Bitcoin Core, but no activation parameters have been chosen.
How do I know whether I'm using Taproot?
Check the address: if it starts with bc1p, it's Taproot. bc1q addresses are the first version of SegWit.
This is not investment advice.



Comments