Decentralization and Safeguards

A chain is as decentralized as the stake that elects its witnesses and funds its proposals. This page gives those figures as they stand, explains where they come from, and lists the protocol rules that limit what any holder can do. It also shows how to recompute the figures yourself, and says what is expected to change them.

The figures on 2026-10-08

1. Down from 85% on 2026-10-05, because the operator's working account pixa matured on 2026-10-05 and does not vote; it holds about 40% of the stake that can vote and is powering down (below).
2. Hardfork 30 lowered the vote dust threshold from 2,500 to 2.5 Pixa Power on 2026-10-07; before it, 21 accounts cleared the line (Voting and Curation).

Where the concentration comes from

The chain is five weeks old, and nearly all of its stake still comes from the genesis allocations (Genesis and Distribution):

• The largest holders received allocations. All nine of the largest holders outside the allocation accounts received their stake from pixa.rex or pixa.team. Accounts that never received allocation stake hold 0.33% of all stake.
• The voters came from the team allocation. The three largest witness voters hold 99.5% of the stake that voted for witnesses. They received it from pixa.team on 2026-09-04. The two voters without stake, tine and eddiespino, hold only delegated stake, which carries no governance weight.
• One holder has a voice it does not use. On 2026-10-05 at 18:12 UTC, 10,884,615 VESTS finished their 30-day wait, 10,384,615 of them belonging to pixa, the operator's working account. pixa holds about 40% of the stake that can vote and casts no governance vote. Its share falls each week as it powers down, 1,153,846 VESTS at a time from 2026-10-10, to nothing by about 2026-12-05.

Who runs the witnesses. Witnesses do not have to say who operates them, and two may sign with the same key: on 2026-11-03 two witnesses that produce blocks, one elected and one taking the timeshare slot in turn, had registered the same block-signing key (Consensus and Block Production). On 2026-10-08, three of the eight scheduled witnesses gave a pixagram.com address as their website, and five gave the address of the public API node each runs (Architecture). One of them, rex, is the account the app uses to open new accounts.

What this means. The protocol keeps the powers apart. The allocations cannot vote, the fund has no keys, and the rules change only through the witnesses. Today, though, a small group of accounts that received the team allocation stands behind each of these powers. Those accounts elect every witness, set the DPF's funding threshold and cast most of the votes that move rewards. The separation exists in the code; it does not yet exist in practice.

Safeguards in the protocol

What the safeguards do not do

• They do not spread stake. Only distribution, rewards and new participants do that.
• They do not separate the people. When the same accounts elect the witnesses and set the funding threshold, the separation of powers depends on those accounts' restraint.
• They do not cover the services most people use. The app at pixagram.com and its account service are run by the operator, Pixa Rex S.A. The chain does not depend on them, and anyone can run a node or build another app (Who Does What).

What should change the figures

This page is updated as the figures move.

Check the figures yourself

This script needs only Python 3. It reads every account from a public node and prints the main figures:

Pythonimport json, urllib.request

NODE = "https://api.pixagram.com"
ALLOCATIONS = {"pixa.rex", "pixa.team"}

def rpc(method, params):
    body = json.dumps({"jsonrpc": "2.0", "method": method, "params": params, "id": 1}).encode()
    request = urllib.request.Request(NODE, body, {"Content-Type": "application/json"})
    return json.load(urllib.request.urlopen(request))["result"]

def vests(text):
    return float(text.split()[0])

names, start = [], ""
while True:
    page = rpc("condenser_api.lookup_accounts", [start, 1000])
    names += [n for n in page if n != start]
    if len(page) < 1000:
        break
    start = page[-1]
accounts = [a for i in range(0, len(names), 100)
            for a in rpc("condenser_api.get_accounts", [names[i:i + 100]])]

total = vests(rpc("condenser_api.get_dynamic_global_properties", [])["total_vesting_shares"])
others = [a for a in accounts if a["name"] not in ALLOCATIONS]
own = sorted((vests(a["vesting_shares"]) for a in others), reverse=True)
matured = {a["name"]: vests(a["vesting_shares"]) - sum(int(d["val"]) for d in a["delayed_votes"]) / 1e6
           for a in others}
voters = [a["name"] for a in others if a["witnesses_voted_for"] > 0 or a["proxy"]]

print(f"accounts:                          {len(accounts)}")
print(f"stake in the allocations:          {100 * (total - sum(own)) / total:.2f}%")
print(f"stake that can vote (matured):     {100 * sum(matured.values()) / total:.2f}%")
print(f"witness voters:                    {len(voters)}")
print(f"  their share of the voting stake: {100 * sum(matured[n] for n in voters) / sum(matured.values()):.2f}%")
print(f"top 5 holders, outside allocations: {100 * sum(own[:5]) / sum(own):.2f}%")

Run on 2026-10-08 at block 979,332, it printed:

accounts:                          100
stake in the allocations:          72.29%
stake that can vote (matured):     27.71%
witness voters:                    7
  their share of the voting stake: 49.55%
top 5 holders, outside allocations: 93.00%

Inherited → changed

Sources

• Hive whitepaper, §III.1.1, "Addressing 51% Attacks", on delayed voting as a response to the events of 2020.
• From STEEM to HIVE to PIXA, with sources for the 2020 events.
• Code, at commit 48f75a2:
  • delayed voting: delayed_voting.cpp:7-122
  • vote expiry: database.cpp:3580-3623
  • the allocation accounts' restriction: database.cpp:2762-2810
• Live chain, read on 2026-10-08: condenser_api.get_accounts for all 100 accounts and condenser_api.get_witnesses_by_vote at block 979,332; database_api.list_witness_votes and list_proposal_votes; the account histories of pixa.rex and pixa.team (2026-10-05). On 2026-11-03, at block 1,723,276: the script above for all 224 accounts, database_api.list_proposal_votes and condenser_api.get_witness_schedule.