Hive Witness Report | Ranking, Voting, Missed Blocks, HBD Interest Changes and More | August 2026
Published on HivePostify by @dalz · Mon Sep 07 2026
How is Hive doing from a witness standpoint? This report looks at the voting support behind the current Top 21, August witness-vote activity, voting concentration, missed blocks, software version readiness, HBD interest-rate settings and price-feed quality. Monthly activity covers the complete UTC month, while ranking, voting power, software and concentration use the latest reproducible state snapshot.
Here we will be looking at: - Number of votes received / removed per witness - Hive Power HP, received / removed per witness - Witness ranking and change through the month - Missed block per witness - HBD interest rate change per witness - Price feed - Software version updates - Witness vote concetration
As mentioned, the data will be for August 2026, with the current-state snapshot taken on the start of September.
Witness Vote Actions
Who gained the most number of new votes, and who lost votes? Here is the chart.
There were 1,034 new approvals in August and 309 removals across all 124 witnesses with vote activity, coming from 253 voter accounts. Compared with July, approvals rose by 98 while removals fell by 166; total actions eased from 1,411 to 1,343. The mix therefore shifted toward additions even though overall transaction activity was slightly lower. The chart screens the complete monthly population and shows the 20 largest gross movements. @stoodkev ranks first with 63 total actions and a net gain of 51, close to its July activity but with one more net approval. @ocd-witness follows with 55 actions and a net gain of 45, while @arcange and @gtg recorded net gains of 37 and 32. Most high-action witnesses finished positive; @stayoutoftherz was the largest removal-heavy name in the chart, with eight additions and fourteen removals. New and removed actions show participation, but not the exact HP change at transaction time.
Witness Hive Power Changes
Individual votes do show the overall support, but what really counts in a stake based system is the mount of HP each witness has gained / lost. Here is the chart for the last month.
Across the full monthly activity set, new approval operations represented approximately 22.04M effective HP, while removal operations represented 25.64M HP. Both gross flows were lower than July, when the comparable estimates were 38.26M and 34.25M HP, and the August balance was approximately -3.60M HP. That reversal shows that a month with more approval transactions can still carry less stake than its removals. This is a snapshot-weighted flow measure, not an exact reconciliation of support. @brianoflondon had the largest positive difference at approximately +1.34M HP, followed by @stresskiller at +1.00M HP and @actifit at +0.99M HP. @threespeak had the largest negative difference at approximately -2.09M HP, followed by @engrave at -1.67M HP and @emrebeyler at -1.39M HP. These larger removal weights explain why the aggregate HP flow was negative despite the strong count of new approvals.
Witness Ranking With the actions above, new support gained and lost, here is the current situation about witness rank and support.
The current ranking is led by @gtg with 104.56M HP in witness support, followed by @ocd-witness with 103.20M HP. Both remain above 100M HP, while ranks 3 through 8 form a relatively tight band from 98.69M to 99.77M HP. At the lower end, @smooth.witness holds rank 20 with 81.22M HP and @howo is rank 21 with 76.20M HP.
The gap between ranks 20 and 21 narrowed to approximately 5.02M HP from 6.45M in the July report, while the full gap between ranks 1 and 21 is 28.36M HP. The number on each bar also shows the current count of witness approvals. It should not be confused with voting weight: one large account can contribute more HP than many small accounts. This month can use the saved 11 August baseline rather than a simple baseline label. @steempeak moved up one place to rank 5, passing @blocktrades. @guiltyparties gained two places to rank 13, while @threespeak fell two to rank 14. @therealwolf and @abit each gained one place; @deathwing and @pharesim each moved down one. The other thirteen positions were unchanged, and no new witness entered the Top 21.
Biggest Governance Movers
Which voter accounts had the most stake behind their August witness-vote changes? The chart below screens every account that added or removed at least one witness approval during August, then ranks the 20 largest by effective HP. The bars measure current account voting weight, while the labels show how many approvals the account added and removed.
The 20 accounts represented approximately 9.90M effective HP, slightly below the 10.16M tracked in July. @ura-soul ranked first at 2.05M HP and removed its approval from @engrave. @acidyo followed at 1.43M HP with two removals, while @stoodkev, at 1.15M HP, removed one approval. @planetoceana changed five choices with 0.84M HP, and @balte recorded eleven actions backed by approximately 0.81M HP. The top three were removal-only accounts, which also helps explain the negative aggregate weight balance.
This ranking identifies large accounts that actively changed witness choices; it does not imply that every operation changed network support by the displayed HP at transaction time. The calculation uses the 4 September effective-HP snapshot, including proxied stake, because historical voter balances and proxy weight are not available for every August operation. Repeated changes by the same account are retained as separate operations, so the figures describe gross governance flow rather than month-end support reconciliation.
Missed Blocks and Witness Reliability Hive scheduled 892,800 three-second slots during August. The chain produced 890,158 blocks and recorded 2,642 producer-missed operations, giving an overall completion rate of approximately 99.704%. This was a modest improvement from July, when 2,808 slots were missed and completion was approximately 99.685%. The decline of 166 missed slots equals a month-over-month improvement of about 5.9%, while total scheduled capacity was unchanged.
The current Top 21 accounted for 170 of the network-wide misses, or 6.4%, down sharply from 669 in the July report. @therealwolf recorded 112 misses from 42,507 scheduled slots, the highest rate in the group at approximately 0.263%. The next-largest count was only eight for @ocd-witness, showing that the current group's misses were highly concentrated in one producer. @howo missed two of 2,480 scheduled slots as the rank 21 backup, a rate of approximately 0.081%. The two busiest disruption days were 11 August with 195 misses and 9 August with 149. Together they contributed 344 misses, or 13.0% of the monthly total. That is much less concentrated than July, when 9 and 10 July generated 787 misses and 28% of that month's total. August still had several elevated days, but there was no two-day shock of the same scale.
Witness HBD Interest-Rate Changes HBD interest changes can be a hot topic. It sets the interest for HBD in savings. So users will pay close attention to this.
Interest policy was much steadier in August. The median for the current Top 21 remained at 10.0% from the start to the end of the month. One witness lowered its setting, twenty ended unchanged and none finished the month with a higher rate. In contrast, nine current Top 21 witnesses had lowered their settings during July as the median moved from 12% to 10%.
The only change came from @pharesim, which moved from 10% to 8%, a reduction of two percentage points. Month-end settings ranged from 6% to 12%: @threespeak and @abit remained at 6%, while @ocd-witness, @arcange and @blocktrades remained at 12%. This followed the much broader set of reductions recorded in July.
Price-Feed Freshness and Dispersion The price feed is an essential parameter for the Hive chain having in mind the dual currency system and the conversions between HBD and Hive. The top 21 witness are feeding the chain with outside prices, serving as oracles.
At the August month-end cutoff, three witnesses had published within 15 minutes, three were between 15 and 30 minutes old, ten were between 30 and 60 minutes old, and five were between one and 24 hours old. No current Top 21 feed was more than 24 hours old, an improvement from the three stale feeds in July. Sixteen of the 21 values were therefore less than one hour old at the cutoff.
The median published value was 0.043 HBD per HIVE. The full range was 0.042 to 0.045, with the highest observation approximately 4.7% above the median and the lowest approximately 2.3% below it. @good-karma supplied the 0.045 observation, while several witnesses were at 0.042. @guiltyparties had the oldest feed at just under four hours, followed closely by @ausbitbank and @good-karma. Freshness improved, although the wider price spread still deserves attention.
Witness Software-Version Readiness At the 4 September snapshot, all 21 witnesses were reporting node version 1.28.7. This completes the convergence visible in the July report, when 18 were on 1.28.7 and @ocd-witness, @threespeak and @emrebeyler were still reporting version 1.28.3. The upgrade gap within the elected set has therefore closed at the version-reporting level.
A 100% share on the newest observed version is a stronger operational signal than last month, although version count alone does not prove hardfork readiness. A node may report the same software version while still differing in operational configuration, failover preparation or hardfork vote. The report should continue to track both the running version and hardfork-version vote, especially before a consensus-changing release.
Witness Vote Concentration Voting concentration eased slightly. The largest direct-voting entity, including stake routed to it through proxies, controls approximately 23.9% of the effective witness-voting stake measured here. The Top 5 account for 42.2%, the Top 10 for 50.9%, the Top 20 for 59.8% and the Top 50 for 70.7%, each a little below the July snapshot. The Top 10 measure fell by roughly 0.6 percentage points month over month.
The largest entity in the snapshot remains @blocktrades, with approximately 32.60M effective HP when its own stake and proxied voting stake are combined. The calculation identified 4,999 direct-voting entities across 63,598 current account-witness vote links. Influence broadens after the largest few entities, but passing 50% within the Top 10 still means a relatively small set of decision paths can affect the whole witness ranking. This calculation groups accounts that vote directly and adds their proxied vote layers. It is a current-state measure taken on 4 September, not a historical August month-end state. The public API result is equivalent to the previous query logic, but the exact timestamp and conversion rate still matter when comparing snapshots. Applying the same public-chain method in later reports will keep changes directly comparable.
All the best @dalz
Tags: #witness#hive#report#aug26#data#ranking#moves