File 3
No Federal Register item cites NHTSA’s 2018 sheet on “94 percent”
In March 2018 NHTSA issued a second sheet on its “94 percent” figure, with the limits of the sample printed again. Six Federal Register items came after it. Three cite the 2015 sheet, three cite neither, and none cites the new one.
Neither sheet prints “human error” or “serious”. NHTSA’s web page said “94 percent of serious crashes are due to human error” from 2017 until a rewrite on January 20, 2022. A final rule printed “roughly 94 percent of all crashes” 102 days later, and states and Congress kept using the figure. Others showed the gap between sheet and sentence years ago. This file credits them and adds the dated trail.
- Status
- Draft
- Dated
- November 6, 2026
- Retrieved
- November 6, 2026
- Responsible
- SOVRINT Inc.
- Corrections
- None
- The claim
- Not established
Draft. AI agents gathered the sources, read them, counted them and wrote this, and no person has checked it yet.
What this file does not say
The claim: not established
94 percent of serious crashes are due to human error.
NHTSA’s sheets count the critical reason assigned to the driver, and say that is not the cause. The material read cannot settle what share of crashes humans cause.
- This file does not say what share of crashes humans cause. Nothing here measures that, and the sheets say they do not either.
- It does not say the figure is false. It records what each text says and leaves true and false to fact-checkers.
- It does not say why any wording changed, or why the 2018 sheet was issued. No document read says, and motive is outside the Method.
- It does not discover the objection. A Stanford post made the point in 2013, writers took it up from 2018 and the AP reported it in January 2022. Prior work lists them.
- It takes no position on automated vehicles, fuel economy or any bill or rule. Lines near them are bare dated quotes.
- It makes no prediction.
Observed
Two sheets, one title
The two sheets share a title and every figure. The second adds 13 sentences and a section on scope and limits, and removes nothing. Neither says why it was issued or that it replaces the other, and NHTSA still serves both, byte for byte as first captured.
| Item | 2015 sheet (archived copy) | 2018 sheet (archived copy) |
|---|---|---|
| Printed | February 2015, DOT HS 812 115 | March 2018, DOT HS 812 506. NHTSA’s catalog created the record on 2018-03-28 |
| The number | “94 percent (±2.2%)† of the crashes” | “94 percent (±2.2%) of the NMVCCS-qualified crashes”, while the summary still says “of the crashes” |
| What it is | “the last event in the crash causal chain” | The same words in the summary |
| On cause | “not intended to be interpreted as the cause of the crash” | The same sentence, and the summary now adds “without suggesting that any of these was cause of the crash”. A new section says “neither suggestive of the cause of the crash nor of a fault” |
| The sample | The survey years and “light vehicles”, and no other condition | 6 a.m. to midnight. EMS dispatched, police at the scene, a light passenger vehicle towed, a police report available |
| Comparing it | Nothing printed | “should not be compared with those from other databases” |
| Human error | Not printed | Not printed |
| Serious, fatal | Not printed | Not printed |
The conditions are not new. The second sheet repeats, word for word, six criteria from NHTSA’s 2008 Report to Congress, and its warning on comparison closely follows that report’s section 5.5. The survey ran from July 3, 2005 to December 31, 2007, and rests on 5,471 crashes, weighted to about 2.2 million.
93.27, 93.47 and 94
- These are shares of weighted survey crashes with a critical reason coded to drivers, and nothing more. The 2008 report prints no such share, and no 93 or 94. Its tables give 2,041,943 of 2,189,166, which is 93.27 percent when divided here. A 2013 Stanford post had already said a 93 could be calculated from its totals.
- The 2015 sheet prints 94. Its own rounded counts, 2,046,000 of 2,189,000, give 93.47 percent. The sheet says percentages use unrounded counts, which it does not print, and the survey data could not be fetched. The ±2.2 is larger than the gap. This is a question for NHTSA, and no finding of error.
- In the sheet’s tables, dividing the rounded counts gives sets that sum to 99, and the printed sets sum to 100. No reason is printed.
Observed
Eighteen lines
One figure, three kinds of claim. The number holds still while the words either side of it move.
- The critical reason, which is the last event in the crash causal chain, was assigned to the driver in 94 percent (±2.2%)† of the crashes.NHTSA, DOT HS 812 115
- ... a factor involved in at least 94 percent of crashes on our roads ... human error.NHTSA Administrator, remarks
- driver error causes 94% of crashesGoogle post
- Sadly, NHTSA estimates 94 percent of highway crashes are a result of human error.NHTSA, vehicle cybersecurity document
- ... too often, 94 percent of the time in fact, fatal crashes can be attributed to a driver’s decision.NHTSA Administrator, remarks
- NHTSA said its research shows that human decisions cause 94 percent of all crashes.Associated Press
- Moreover, NHTSA's data suggest that 94% of crashes can be tied to a human choice or behavior.Federal Register, NHTSA notice
- NHTSA (National Highway Traffic Safety Administration) found that 94 percent of those deaths were the result of human error ...Ranking member, Senate subcommittee
- ... rooted in one critical and tragic fact: 94 percent of serious crashes are due to human error.NHTSA website
- The critical reason was assigned to the driver in an estimated 94 percent (±2.2%) of the NMVCCS-qualified crashes (Table 1).NHTSA, DOT HS 812 506
- WHEREAS, roughly 94 percent of automobile crashes are caused by human error ...Massachusetts agreement with 14 cities and towns
- ... human error was linked to 94 percent of serious injury crashes.City of Charlotte, Vision Zero Action Plan
- According to the Department of Transportation, 94 percent of serious crashes are due to driver error.Ranking member, House subcommittee
- ... the effect of human error in crash causation, a contributing factor in roughly 94 percent of all crashes ...Federal Register, NHTSA final rule
- 94 percent of crashes are due to human error ...Federal Highway Administration report
- ... a whopping 94 percent of serious crashes--are due to driver factors ...Chair, House subcommittee
- Regardless of the researching entity, collectively, these studies have determined that driver errors cause at least 94 percent of traffic collisions.Minnesota Office of Traffic Safety, legislative report
- Approximately 94 percent of all traffic crashes are attributed to human error.Senate committee hearing page
The Method says two texts carry one narrative when they make the same claim about the same subject. These lines do not all do that, and that is the finding, so this file follows one figure and sorts each line by what it claims. Coded: the critical reason assigned to the driver, which is the measure. Involvement: tied to, a factor in, linked to. Cause: due to, caused by, a result of. The sorting is this file’s reading, and “traced to” in the rule text of 2017 could fall either way.
Eighteen of the lines read, chosen to show each change once. Each row opens its entry below. Press and company lines are short excerpts. Footnote marks are left out, and “...” marks a cut. Officeholders are shown by office. The left column says what the 94 is a share of in that line.
- Coded
- 2 lines
- Involvement
- 4 lines
- Cause
- 12 lines
Observed
How the wording moved
Each date is the earliest sighting found of that change. Most changes happened in NHTSA’s own texts, and within two years of the sheet.
The range goes
From 94 percent (±2.2%)
To 94 percent
The first sighting
the human 94 percent of the time
NHTSA Administrator, remarks (archived copy)
One month after the sheet. In federal text the range returns only in one footnote, in 2018.
Assigned becomes causes
From the critical reason was assigned to the driver
To driver error causes
The first sighting
driver error causes 94% of crashes
NHTSA’s own texts follow: “a result of” in July 2015, “caused by” in its strategic plan of October 2016.
All crashes become fatal crashes
From the crashes in the survey
To fatal crashes
The first sighting
That’s the percentage of fatal crashes in which the critical reason ... can be traced to human behavior.
NHTSA Administrator, remarks (archived copy)
Neither sheet prints the word fatal.
A figure becomes a ceiling
From 94 percent
To as much as, as many as, up to
The first sighting
as much as 94 percent of fatal roadway crashes
NHTSA workshop summary (archived copy)
A floor runs beside it: “at least” from 2015, “more than” in Missouri’s report for 2016, “at least” again in Minnesota’s of 2024.
Crashes become serious crashes
From crashes, or fatal crashes
To serious crashes
The first sighting
94 percent of serious crashes are due to human error.
NHTSA budget for fiscal 2018 (archived copy)
Dated by the file’s name and date, and no copy of it was captured before 2018-12-30. The Secretary of Transportation said it on 2017-09-12, the day before the first page capture. The word has an older history. See “Where serious comes from”.
The agency goes back to critical reason
From due to human error
To the critical reason
The first sighting
human error or choice are estimated to be the critical reason in 94 percent of crashes
Federal Register, NHTSA (archived copy)
The wording does not run one way. NHTSA’s budget of February 2018 also says “critical reason”, about “serious crashes studied in a crash causation survey”, while the web page kept its sentence.
One survey becomes these studies
From NHTSA’s survey
To these studies
The first sighting
these studies have determined that driver errors cause at least 94% of traffic collisions
Minnesota Office of Traffic Safety (archived copy)
Repeated in the state’s legislative report of 2025-02-27.
Observed
The page, by capture
NHTSA’s automated vehicles page, read across 870 changes in the archive’s index from 2016 to 2026. Dates are the first and last capture of each state, in UTC. They bound a change and do not date it.
| First capture | Last capture | Sentence | Tile |
|---|---|---|---|
| 2016-09-23 | 2016-12-14 | 94 percent of crashes tied to a human choice or error | None |
| 2016-12-23 | 2017-06-16 | we can tie 94 percent of crashes to human choice | 94 % Crashes due to human error |
| 2017-09-13 | 2017-09-23 | 94 percent of serious crashes are due to human error | 94% of Serious Crashes Due to Human Error |
| 2017-10-26 | 2018-01-05 | 94 percent of serious crashes are due to dangerous choices or errors people make behind the wheel | 94% of Serious Crashes Due to Human Error |
| 2018-01-25 | 2019-06-16 | Of all serious motor vehicle crashes, 94 percent are due to human error or choices. | None |
| 2019-06-26 | 2022-01-20 | 94% of serious crashes are due to human error | None |
| 2022-01-20 | 2026-10-28 | No 94. A new page: “reduce human errors and the resulting crashes, injuries, and economic tolls” | None |
The change of January 2022 fell between captures 16 hours 29 minutes apart, and rewrote the whole page. The AP reported two days earlier that NHTSA would update the wording, “to address that characterization of the data as well as provide additional information.” No NHTSA notice of the change was found.
Observed
The series
Nine items in 984 weeks, and no week holds more than one. This is a series of the figure, not of the claim: no item pairs 94 with “serious” or with “due to human error”. It is a set of dated marks and will not make a line. The cause claim sits outside this source, on web pages and in hearings.
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
Prints the number
Cites the sheet, no number
The 2018 sheet, by NHTSA’s catalog date
| Date | Document | What it says | Cites |
|---|---|---|---|
| 2016-22993 Request for comment | 94% of crashes can be tied to a human choice or behavior | 2015 sheet | |
| 2017-01542 Denial of a petition | 94 percent of vehicle crashes can be traced to human choices ... or error | None found | |
| 2018-00671 Request for comment | the critical reason in 94 percent of crashes | 2015 sheet, reprinting its summary with the range | |
| 2018-16820 Proposed rule | a contributing factor in roughly 94% of all crashes | Neither sheet | |
| 2019-11032 Advance notice | No number. “critical factors behind the occurrence of a large number of crashes” | 2015 sheet | |
| 2020-25930 Advance notice | No number. “the substantial role that human factors ... play in contributing to crashes” | 2015 sheet | |
| 2021-17496 Proposed rule | a contributing factor in roughly 94% of all crashes | Neither sheet | |
| 2022-07200 Final rule | a contributing factor in roughly 94 percent of all crashes | Neither sheet | |
| 2024-29862 Final rule | No number. The sheet is cited at “police-accident reports”, in a sentence on studies of driver error | 2015 sheet |
How it was counted
The Federal Register. Distinct items per calendar week, by publication date, from 2008-01-01 to 2026-11-06. A true item states this statistic about crashes, or cites the sheet. 67 written searches gave 3,436 documents, and the full text of all of them was fetched. 89 search counts are logged below. A second agent reran all 89 with matching counts.
2021-17496 and 2022-07200 are one sentence in one rule series, changing only “94%” to “94 percent”, so they fold. 2018-16820 shares the clause in a different rule series. Folding that too gives 7 items, 4 printing the number.
Only the footnote of 2018-00671 keeps the range and the survey years. It reprints the 2015 summary and leaves out the sheet’s “not intended” sentence and the sample conditions. “Proposed Rule” is a Federal Register section label: four of the items filed there are other kinds of document, as the table shows.
- Weeks searched
- 984
- Items
- 9, in 9 weeks
- Copies folded
- 8
- Printing the number
- 6, or 5 folded
- After the 2018 sheet
- 6
- Citing the 2018 sheet
- 0
- Pairing 94 with serious
- 0
- Printing “not intended”
- 0
Observed
Where “serious” comes from
Neither sheet prints the word, and no NHTSA document read defines it. It did not arrive from outside NHTSA, though.
Eleven sets of remarks by NHTSA’s Administrator and Deputy Administrator, from 2011-03-18 to 2012-10-08, name the survey and say “about 95 percent of serious crashes”. The first reads: “in about 95 percent of serious crashes driver error was attributed to the event that precipitated the crash”.
The same office said “about 90 percent” fifteen days after the last of those. A 93 appears beside the critical reason in a news item of 2012 and on NHTSA staff slides of 2013. All 148 NHTSA speech PDFs in the archive index up to 2015 were read, and none prints a 93 or 94 share before the sheet. These figures sit side by side. No text read says how they relate, so 95 and 94 are not shown here as one number moving.
Assessed: NHTSA could defend the word as shorthand for the sample, which required EMS to be sent. Two printed criteria have nothing to do with severity, though: the hours of 6 a.m. to midnight, and a light passenger vehicle towed. A crash at 2 a.m. does not qualify however severe it is. So on this reading the word does not stand for the sample. The weight of this file sits on the verb, due to, against the sheet’s “not intended to be interpreted as the cause”.
Remarks of 2011-03-18, remarks of 2012-10-08, both archived. NHTSA slides of 2013 (archived copy).
Observed
After the page changed
The NHTSA page changed on 2022-01-20. The figure stayed in use elsewhere.
- Federal Register
- The final rule of 2022-05-02 repeats its proposal’s sentence. Nothing later prints the number.
- Other federal text
- Two FHWA reports of late 2022. No new federal text after 2022-10 was found. Older documents stayed online, which is a different thing from saying it again.
- Congress
- A ranking member on 2022-02-02 and a subcommittee chair on 2023-09-13, plus a Senate committee page of 2026. In the 15 hearing records read, from 2016 on, the minority side used it at 8 and the majority side at 7, under four administrations. This is what search surfaced, not a census.
- State pages
- Live on 2026-11-06 in Florida, Utah, Vermont, California and Missouri. Missouri’s may be a separate lineage, and it is left out of every count here.
- Local
- A sample of 14 city, county and regional plans from about 2023 to 2025 printed it in none, and that sample is weak. Six local project summaries in Michigan’s fiscal 2025 application do, four of them on one template. Santa Ana’s 2024 draft plan rejects it under the heading “The 94% Myth”.
State pages: Florida (archived copy), Utah (archived copy), Vermont (archived copy), California (archived copy), Missouri (archived copy). A company page, by way of example: Waymo FAQ (archived copy).
| Set | Carry it | Read | Which |
|---|---|---|---|
| Fiscal 2021 plans | 4 | 57 | Iowa, Maine, Missouri, Washington |
| Fiscal 2023 plans | 1 | 57 | Washington |
| Fiscal 2024–26 plans | 3 | 50 | Minnesota, Pennsylvania, Washington |
| Fiscal 2025 applications | 1 | 56 | Michigan, in local project summaries |
| Fiscal 2026 applications | 0 | 56 | None |
Every plan or application in five sets NHTSA lists, read from archived copies. Each file was searched for the number near the words human, driver, error or behavior, and every match read. This covers one document type. Zero in fiscal 2026 does not show state use has ended.
Observed
The lines in full
Earliest observed source does not establish original authorship.
2015-02
The critical reason, which is the last event in the crash causal chain, was assigned to the driver in 94 percent (±2.2%)† of the crashes.
2015-03-14
... a factor involved in at least 94 percent of crashes on our roads ... human error.
2015-05-11
driver error causes 94% of crashes
2015-07-20
Sadly, NHTSA estimates 94 percent of highway crashes are a result of human error.
2015-09-22
... too often, 94 percent of the time in fact, fatal crashes can be attributed to a driver’s decision.
2015-11-24
NHTSA said its research shows that human decisions cause 94 percent of all crashes.
2016-09-23
Moreover, NHTSA's data suggest that 94% of crashes can be tied to a human choice or behavior.
2016-11-16
NHTSA (National Highway Traffic Safety Administration) found that 94 percent of those deaths were the result of human error ...
2017-09-13
... rooted in one critical and tragic fact: 94 percent of serious crashes are due to human error.
2018-03
The critical reason was assigned to the driver in an estimated 94 percent (±2.2%) of the NMVCCS-qualified crashes (Table 1).
2018-09-21
WHEREAS, roughly 94 percent of automobile crashes are caused by human error ...
Massachusetts agreement with 14 cities and towns (archived copy)
2019-03-22
... human error was linked to 94 percent of serious injury crashes.
2022-02-02
According to the Department of Transportation, 94 percent of serious crashes are due to driver error.
2022-05-02
... the effect of human error in crash causation, a contributing factor in roughly 94 percent of all crashes ...
2022-10
94 percent of crashes are due to human error ...
2023-09-13
... a whopping 94 percent of serious crashes--are due to driver factors ...
2025-02-27
Regardless of the researching entity, collectively, these studies have determined that driver errors cause at least 94 percent of traffic collisions.
Minnesota Office of Traffic Safety, legislative report (archived copy)
2026-02-04
Approximately 94 percent of all traffic crashes are attributed to human error.
Observed
Prior work
The gap between the sheet and the sentence was found by others first. This file builds on their work.
2013-12-18
Stanford Center for Internet and Society blog (archived copy)
Said the 2008 report’s critical reason may not be the cause, that a 93 is absent from the report though calculable from its totals, and listed the older studies. The earliest of this kind found. Read from a capture of 2014, because the live post was later updated.
2018-06-05
Phil Koopman, Safe Autonomy blog (archived copy)
Made the same point from the 2015 sheet, with the 2008 source and the age of the data.
2018-08-03
Frontier Group (archived copy)
Set NHTSA’s web sentence beside the sheet.
2018-12-19
Don Kostelec, Kostelec Planning (archived copy)
Showed the agency restating itself, and in Streetsblog in 2020 listed who else used the figure.
2021-11-26
David Zipper, The Atlantic (archived copy)
Set the sheet’s caveat against the web sentence and named carriers of the figure.
2022-01-18
Associated Press (archived copy)
Reported the NTSB chair’s objection, and NHTSA saying it would update its website wording.
2022-01-20
David Zipper, post carried by Streetsblog (archived copy)
Reported publicly that the website had changed, on the day it did.
2026-10-07
Printed the sample conditions, and said the survey defines no category of serious crashes.
Assessed
What this file adds
- The March 2018 sheet read against the 2015 sheet. TRID lists it as a separate publication, and no text read calls it a reissue or says it differs.
- The bound on when “serious” reached the page, 2017-06-16 to 2017-09-13, and the removal to within 17 hours. The date of the removal was already public.
- 93.47 from the sheet’s own counts against a printed 94, kept as a note and a question for NHTSA.
- The Federal Register trail of nine items, including the final rule of 2022.
- NHTSA’s own “about 95 percent of serious crashes” in 2011 and 2012.
- The census of filed state plans.
Assessed
How far that goes
“Adds” means not found in the 15 works term-checked. That is absence of evidence, and a person should test it before anyone relies on it.
Observed
Copies
Repetition is not corroboration, so these count once each.
2015-11-24
NHTSA said its research shows that human decisions cause 94 percent of all crashes.
One source. An AP story on two sites checked.
2022-01-18
94% of serious crashes are solely due to driver error
One source. An AP story on five sites checked: Transport Topics, WJLA, Sentinel Colorado, Insurance Journal, Claims Journal.
2021-09-03
a contributing factor in roughly 94 percent of all crashes
One sentence, carried from the proposal into the final rule of 2022-05-02.
2024-11-20
94 percent of all crashes are the result of driver error
One template in Michigan’s application, on four local summaries and one state summary.
Limits
What went unread, and how far a “none found” can be trusted.
All 10 limits
- NHTSA has not been asked. No NHTSA notice on the 2018 sheet or the 2022 change was found. The AP report is the only NHTSA statement found.
- nhtsa.gov, transportation.gov, fhwa.dot.gov, commerce.senate.gov and other federal hosts refused automated fetches. Those lines rest on archived copies, and the state of those live pages today was not seen.
- Capture dates bound a change. They do not date it. No capture of the page exists between 2017-06-16 and 2017-09-13, or within the 16 hours 29 minutes on 2022-01-20.
- The survey data could not be fetched, so no 93 or 94 could be reproduced from case data.
- The Federal Register count covers what its index returns for the written searches. 805 documents were screened by machine alone. 201 were picked by a keyword rule and rejected, and a sample of them was read. A second agent found no missed item in 713 further documents.
- GovInfo’s search refused the shared key, so hearings were found by web search and read by address. No count of hearings is claimed.
- An earliest sighting is what these searches found. Google Books refused, and congressional full text before 2015 was not searched. It does not establish authorship.
- State annual reports, strategic highway safety plans and most web pages are outside the census. Image-only pages in any filed plan went unsearched.
- Books were not read, including Killed by a Traffic Engineer (2024), reported to discuss this figure.
- The Missouri page’s 33 and 13 sit near the 1977 Tri-Level study, but its 94 does not match that study’s 92.6, and it names no study. Any link is inference.
Not established
Other explanations
The pattern has more than one possible reading. Each of these fits what was read, and none says anyone meant anything.
Not established
Ordinary compression.
Each restatement shortens a technical phrase, and later writers copy the short form. Wording moves in many offices at once, and in both directions.
Not established
Different offices, different texts.
The sheets come from statisticians, the web page from another office and the rule text from rule writers. The 2022 footnote repeats a sentence from 2021, and that clause is in a 2018 proposal from another rule series.
Not established
The figure took over an older idea.
“Due to driver error” sits beside 90 percent in NHTSA remarks of 2012, and the 1977 Tri-Level study says about 90 to 93. So the words may be older than the number, and 94 may simply have been the newest figure to hand.
Not established
95, 93 and 94 are one table on different bases.
2,041,943 is 95.43 percent of the 2,139,624 crashes with a known reason, and 93.27 percent of all 2,189,166. This is this file’s arithmetic. No text read says so.
Not established
The 2018 sheet was not treated as a replacement.
It has the same title, says nothing of replacing the first, and both are still served. The 2015 sheet was missing from captures only for some months in 2023 and 2024.
Not established
Serious is shorthand for the sample.
The sample required EMS dispatch, NHTSA leaders used the word in 2011, and NHTSA’s budget of 2018 says “serious crashes studied in a crash causation survey”. The hours and the towed light vehicle are not about severity, and no document read defines the word.
Not established
The search missed a citation.
Two agents searched separately and found none, and “812 506” returns nothing. The index could still miss a document.
Status of the claims
- 94 percent of serious crashes are due to human error.
- Not established
- NHTSA’s sheets count the critical reason assigned to the driver, and say that is not the cause. The material read cannot settle what share of crashes humans cause.
- The critical reason was assigned to the driver in an estimated 94 percent (±2.2%) of the NMVCCS-qualified crashes.
- Observed
- Printed in the March 2018 sheet. This describes the text. How the percentage was reached could not be recomputed, because the unrounded counts are not printed and the survey data could not be fetched.
What the status words cover
- Observed
- The text was seen at its source or in an archived copy, saying what is quoted. The claim inside it may still be wrong.
- Assessed
- This file’s own reading of what was observed.
- Not established
- The material cannot settle the question.
Made by agents, unchecked
AI agents gathered the sources, read them, counted them and drafted this file, using scripts. A second set of agents went back to the sources and tried to refute each quote and date. No person has yet checked it or read a source. No SOVRINT software produced it.
Sources, search, counting and tools
- Sources
- NHTSA’s sheets and 2008 report, the Federal Register API, archived captures of NHTSA’s page, 148 archived NHTSA speech PDFs, hearing records on GovInfo, filed state plans, and the pages linked in this file.
- Search
- Every Federal Register search is listed below with its count. Web searches for carriers and prior work ran through a web search tool and their strings were not all logged, so a “none found” resting on them is weaker for that.
- Counting
- Distinct items per calendar week from the Federal Register, with a fold shown beside the raw count.
- Tools
- curl, pdftotext, pdfinfo, the Federal Register public API and the Wayback index, with scripts to extract windows of text. A local copy of each source was kept with its hash.
- Refusals
- Pages that returned 403, 404 or 429 were recorded and not worked around. No new archive captures were triggered.
- Snapshot
- All sources were read on November 6, 2026. Archive addresses are given where a capture exists.
- Left out of scope
- What share of crashes humans cause, the safety of automated vehicles, the merit of any rule or bill, and anyone’s intent. Private citizens’ testimony was left out, and elected officials are shown by office.
The 89 logged Federal Register counts
- "94 percent" crashes: 44
- "94%" crashes: 614
- "94 percent" "human": 588
- "human error" crashes: 84
- "critical reason": 21
- "National Motor Vehicle Crash Causation Survey": 18
- NMVCCS: 16
- "812115": 3
- "812 115": 2
- "812506": 0
- "812 506": 0
- "19 of every 20": 1
- "ninety-four percent" crashes: 3
- "94 percent" crash: 44
- "94 percent" "human error": 22
- "94%" "human error": 209
- "94 percent" "driver error": 0
- "94%" "driver error": 10
- "94 percent" collisions: 70
- "94%" collisions: 586
- "94 percent" "human choice": 1
- "94 percent" "automated vehicles": 5
- "94%" "automated vehicles": 32
- "94 percent" "human factors": 33
- "94%" "human factors": 225
- "94 percent" "motor vehicle" driver: 68
- "94%" "motor vehicle" driver: 755
- "critical reasons": 12
- "Critical Reasons for Crashes": 4
- "crash causation survey": 18
- "Crash Causation" NHTSA: 62
- "critical pre-crash event": 1
- "DOT HS 812 115": 2
- "HS 812 115": 2
- "HS 812 506": 0
- "HS 812115": 0
- "HS 812506": 0
- "nineteen of every twenty": 0
- "19 out of 20" crashes: 0
- "human choice" crashes: 2
- "human choices" crashes: 2
- "ninety-four percent" crash: 3
- "ninety four percent" crashes: 3
- "94 per cent" crashes: 0
- "94 percent of serious": 0
- "94% of serious": 0
- "94 percent of crashes": 1
- "94% of crashes": 1
- "94 percent of all": 30
- "94% of all": 16
- "due to human error" crashes: 10
- "attributed to human error" crashes: 4
- "caused by human error" crashes: 11
- "90 percent" "human error" crashes: 28
- "nine out of ten" crashes: 2
- "9 out of 10" crashes: 0
- "94 percent" accident: 105
- "94%" accident [agency: transportation-department]: 398
- "automated driving systems": 128
- "automated vehicles": 110
- "human error" [agency: transportation-department]: 132
- "94%" [agency: national-highway-traffic-safety-administration]: 219
- "94%" [agency: federal-motor-carrier-safety-administration]: 250
- "94%" [agency: federal-highway-administration]: 168
- "94%" [agency: transportation-department]: 1602
- "human error" [agency: national-transportation-safety-board]: 2
- "94%" crash [agency: national-transportation-safety-board]: 1
- "effect of human error": 3
- "contributing factor in roughly": 3
- "tied to a human choice": 1
- "traced to human choices": 1
- "critical factors behind": 2
- "not intended to be interpreted": 12
- "assigned to the driver": 3
- "human error or choice": 1
- "human error and choice": 2
- "last event in the crash causal chain": 1
- "overwhelming majority of accidents": 5
- "overwhelming majority of crashes": 3
- "interpreted as the cause": 0
- "assignment of the fault": 0
- "assignment of fault": 2
- "cause of the crash" "critical reason": 0
- "94 percent" "human error" automated: 20
- "due to human error" "94 percent": 1
- "due to human error" "94%": 15
- "serious crashes" "94 percent": 1
- "serious crashes" "94%": 9
- "serious crashes" "human error": 3
Corrections
None so far.