The Los Angeles City Council stopped agreeing

The Los Angeles City Council has recorded 124,331 floor votes since July 1, 2003, and for most of that time almost none of them drew a no. From 2014 through 2019, 0.85 percent of roll calls had even one no vote. Since January 2025 the share is 6.45 percent, the highest in the record, and it does not come from four members alone.

No votes are 0.33 percent of all member-votes

Across 1,864,887 member-votes, excluding the 28,319 rows where a seat was vacant, the council voted 82.07 percent yes, 17.60 percent absent, and 0.33 percent no. That ordering means a Los Angeles councilmember is 53 times more likely to be recorded absent than to vote no. Absence, rather than dissent, is the ordinary way of not supporting an item.

Only 1.85 percent of all 124,331 roll calls drew even a single no vote. In 23 years, exactly two items were voted down with zero ayes: a planning department item on a Venice Boulevard project in August 2017, defeated 0 to 13, and an alley vacation off Santa Monica Boulevard in April 2020, defeated 0 to 15. Both were land use in Council District 11. That near-unanimous pattern is the baseline for everything that follows, and in the last few years it has changed.

The rise

The record starts July 1, 2003 and ends August 12, 2026, so the first and last rows are partial years. That means every number touching 2026 is a partial-year number.

The 2014–2019 baseline isn't a cherry-picked trough. The 2003–2009 era ran at 0.90 percent, essentially the same, which gives two stretches a decade apart both landing near nine roll calls in a thousand. Today it's about 65 in a thousand. 2025 is the single most divided year in the record at 6.60 percent, ahead of the partial 2026 at 6.19 percent and 2024 at 5.35 percent. No year before 2020 came close; the pre-2020 high was 3.24 percent, in 2010.

The council takes fewer votes than it used to

Roll-call volume roughly halved over this period. The council recorded 8,550 votes in 2007 and 4,179 in 2025, with 2,519 so far in 2026. Per meeting day it fell from 68.9 to 36.7. A council that holds the same number of disputed roll calls while taking half as many votes will show a doubled dissent share without anyone changing their behavior. A rate per member-vote is less sensitive to that, though it still moves if the mix of unanimous items changes, and by that measure votes of no per 1,000 member-votes rose from 1.10 in 2014–2019 to 12.08 in 2025–2026, an 11.0-fold rise against the 7.6-fold rise in the share. The absolute counts run the same way. There were 270 disputed roll calls in six years from 2014 through 2019. There have been 432 in the nineteen months since January 2025.

Subtracting the left bloc

The most visible change on the council is the arrival of a left bloc: Nithya Raman in 2020, Eunisses Hernandez and Hugo Soto-Martinez in 2022, Ysabel Jurado in 2024. Hernandez and Soto-Martinez are the two most frequent no voters in the entire 23 year record, with 467 and 415 no votes.

Subtracting them shows how much of the rise they account for. Across the 6,698 roll calls of 2025 and 2026, ignoring every no vote cast by Hernandez and Soto-Martinez drops the share with a surviving no from 6.45 percent to 5.29 percent. Ignoring all four of Hernandez, Soto-Martinez, Jurado and Raman drops it to 3.33 percent, which is still 3.9 times the 2014–2019 baseline. More than half of today's disputed items, 223 of 432, include a no vote from someone other than those four.

Dissent also rose among the five members who sat on this council in both windows, the same people in the same seats a decade apart.

Every one of them dissents more, by between 4.7 and 34-fold. Rodriguez, Lee and Harris-Dawson first appear in the record in 2017, 2019 and 2015, so their baselines rest on partial service. Blumenfield and Price were present for all 31,904 roll calls of the baseline window, and both rose about sevenfold.

The medians move the same way. The median councilmember cast votes of no at a rate of 0.96 per 1,000 in 2014–2019 and casts them at 8.21 per 1,000 today, or 6.57 excluding those four. Curren Price is the least dissenting member of the current council at 3.73 per 1,000, and that rate would have made him the second most dissenting member of the 2014–2019 council, behind only Bernard Parks at 3.82.

All 15 current members have voted no at least once since January 2025, and the average disputed item now draws 2.81 no votes, up from 1.93. Disputed items have become both more common and more widely joined.

What they are dividing over

The table below ranks the ten most contested council files of 2025 and 2026 by the number of no votes cast.

The largest cluster of no votes sits in a single file. On June 17, 2026, the Charter Reform Commission amendments produced 47 separate roll calls in one day, 22 of which drew a no vote, and 21 of those 22 included a dissenter from outside the four. The council was voting amendment by amendment on the rules governing its own structure, and the dissent was spread across the body rather than confined to one bloc.

Three sensitivity tests, each measured against the 0.85 percent baseline of 2014–2019. Removing that entire June 17 meeting puts the 2025–2026 share at 6.16 percent. Removing the meeting and the four left bloc members puts it at 3.01 percent. Removing the four members and the five most contested council files puts it at 2.55 percent, still 3.0 times the baseline.

A third, independent measure moves the same way. Roll calls where votes of no outnumbered ayes went from 0.053 percent of roll calls in 2014–2019 to 0.388 percent in 2025–2026, so items are now defeated on the floor about seven times as often.

Method

There's no bulk source for how the council votes. Those votes aren't on data.lacity.org, and the city's Legistar instance returns a configuration error instead of data. What exists is one web page per vote inside the City Clerk's Council Votes Verification System, which has no export and no API, so we enumerated all 132,696 vote ids in the system, parsed every live page, and built the file ourselves. As far as we can find, this is the first time the record has been published as a dataset.

The dissent share is the share of recorded floor roll calls in which at least one councilmember was recorded voting NO, computed from member-level rows rather than the page header tally. It excludes vacant seats, since 28,319 member rows carry a VACANT name string, 28,110 of them recorded absent, and leaving them in pushes the citywide absence rate from 17.60 to 18.84 percent while diluting every other share. It also merges Kevin de Leon, whom the clerk's system carries under two member ids and two spellings, KEVIN DELEON and KEVIN DE LEON. Without the merge he shows up as two councilmembers.

The artifact we ruled out is the meeting type field, which reads SPECIAL on all 124,331 pages. That's vestigial in the source system, not a parsing failure, so any analysis splitting regular from special sessions here is measuring nothing, and we dropped the column.

There are two limits worth stating. The clerk's system has no abstention token, so a member who recuses or steps off the floor is recorded absent alongside one who never came to work, and the three can't be separated from this source. And the header tallies disagree with the member rows on absences in 276 roll calls, 0.22 percent, 275 of them in 2005. We publish both.