Talk:List of rail accidents/Criteria for inclusion

From Wikipedia, the free encyclopedia

You have new messages (last change).

Contents

[edit] Early comments

There doesn't appear to be any criteria for what "accidents" get included in this list. I believe, for example, that level crossing accidents (perhaps unless they cause a derailment or worse) should not be included. Also people falling off trains should not be included. Neither are really railway accidents; i.e. due to some fault (human or otherwise) of the railway itself, nor do they have significant consequences to the railway (as a terrorist attack could have). Philip J. Rayment 09:12, 22 November 2006 (UTC)

User:Nutmegger deleted my recent additions, he disqualified the accidents because no lives were lost, yet he enhanced the edit about a terrorist bomb attack. I took the liberty of reverting my additions (and that of a German accident which he deleted as well).
To my humble opinion, a rail accident that should be included in this list, is something that caused enough damage to disrupt the normal train circulation and has been caused by either a mechanical failure to the rails or train (e.g. Eschede, Germany), or a human error (Signal passed at danger, or a signalling failure). In other words: the accident should be rail-related. A terrorist attack should not be included. I am composing a list of accidents in The Netherlands (because no page like this one exists at nl.wikipedia), but I should save my energy until it is clear what should be on this list, and what should be left out. I am only a humble non-native English speaker, so if you cannot understand me because of my heavy accent, that's fine with me ;-) I am also known as nl:Gebruiker:Brinkie --Brinkie 09:38, 23 November 2006 (UTC)

[edit] Thryduulf's initial suggestions

I agree we need to agree a criteria for inclusion. My suggestions are: Part A:

  1. If the accident is caused by suicide, vandalism or trespass, or factors completely unrelated to rail operations, then it should NOT be included unless it receives significant coverage in non-specialist national (national list only) or international media. i.e Queen of the Sea train disaster does belong on this list.
  2. If the only fatalities/serious injuries are people falling off/out of trains not involved in a collision/derailment/etc then it should NOT be included.
  3. If the accident is caused by terrorism, then it should NOT be included unless it has significant long term effects on railway operations other than security.

Part B:

  1. At least one person (passenger or crew) on the train is killed, or five people receive serious injuries
  2. A train hits a road vehicle carrying 6 or more people, excluding people carries and similar vehicles classed as cars.
  3. The accident causes significant infrastructure damage. Significant means it takes at least 24 hours to repair.
  4. Modification to all rolling stock of the same/very similar (e.g. British Rail Class 220/221/222) type is required as a result
  5. Three or more separate trains are involved
  6. The accident leads to a closure of a significant portion of the system, including parts not direct affected, lasting at least three days. (e.g. Chancery Lane derailment)
  7. A statutory body (e.g. Rail Accident Investigation Branch, Rail Safety and Standards Board) produces a report into the accident. This does not include reports by police or Health and Safety Executive or similar bodies.
  8. There is a Wikipedia article specifically about the accident (i.e. independent of an article about the system or location). If the article is deleted, then it does not qualify under this criteria.
  9. The accident receives significant coverage in non-specialist regional or national media.
  10. The accident receives extensive coverage in national specialist media.
  11. The accident is otherwise historically significant, e.g. it is the first.

Part C:

  • If any of the following are true then it should be included on this, international list. and the relevant national list.
  1. At least ten people on the train (passengers or crew) are killed (50% if there are less than 10 on board), or twenty people receive serious injuries (75% if there are less than 20 on board).
  2. A train hits a road vehicle carrying 10 or more people, at least 5 of whom are killed or the total death toll in the accident is at least 10.
  3. Modification to a significant proportion of a system's rolling stock, not just of the type involved.
  4. The accident leads to significant changes to the system's infrastructure (e.g. such as Selby rail crash)
  5. A national private or public inquiry is held into the accident
  6. The accident receives significant coverage in international non-specialist media
  7. The accident receives extensive coverage in the national non-specialist media of a medium-large country.
    • For television and radio, the coverage must last more than two days.
    • For printed media, the coverage must feature in at least three editions of daily media, and/or at least two editions of media published less frequently.
  8. The accident is otherwise internationally historically notable (e.g. first ever train accident on a continent, led to international safety regulation changes, first ever accident on that type of system (e.g. funicular, maglev)) AND qualifies for an entry on the relevant national list.

The "national" lists can be for single countries (e.g. UK, USA) or groups of countries/regions where we have fewer accidents (e.g. eastern Europe). To fully cover the terrorist incidents, perhaps we should create a List of terrorist incidents on rail transport, added as a prominent see also. Thryduulf 00:26, 25 November 2006 (UTC)


  • I agree with almost all of the above suggestions, my only concerns being with part A. If an incident involves suicide, trespass etc, but also fulfills one or more of the stipulations of parts B or C, then should it be included in either the international list or not? Whether the incident is the result of a technical fault or not, it is still a rail accident and perhaps should be included e.g. vandals leaving debris on a track resulting in a crash killing three people, or a tree falls and derails a train killing two - are these examples eligible for the national lists or not?

Secondly, there is at least one incident on the list of sixty people being swept from the roof of a train by a low bridge. I agree that individual falls from trains are not-notable enough, but should an exception be made for incidents of this magnitude despite their not involving an actual crash? Thirdly, I agree that terrorist incidents need not be directly included on the list, but a question is raised over deliberate derailments (such as the Rafiganj train disaster). This incident falls between the two lists, and a clarification should be made as to which it and others like it belong to or even if it should be on both.

In addition, a seperate list for terrorist incidents on railways should be a high prioirty for setting up, and my intial suggestion is that national lists be created or maintained for the USA, UK, Australia, India, Germany and possibly France, Spain, Russia and China with others being coralled into regional lists (Africa, South-East Asia, South America etc) and then branched out as and when it becomes necessary. There may well be other countries which require a page of their own, this is just a selection of ideas. Otherwise, I think the proposals above are very sound, and very necessary to remove the volume of less important incidents which are beginning to clutter the list.--Jackyd101 02:12, 25 November 2006 (UTC)

The suggestions sound reasonable. However, these should be implemented along with a separate list for terrorist incidents on railways and if possible, be managed by same wikiproject(s). --Gurubrahma 10:03, 25 November 2006 (UTC)
  • In Australian cases the criterion would be much simpler: it merits inclusion if and only if it is a gazetted disaster, which would be determined by whether or not it appears in the Emergency Management Australia Disasters Database, "the primary Australian Government database containing records of all natural and non-natural disasters within Australia." This criterion can only be used in Australian cases, so by all means set out a more general set of criteria. But I wouldn't like to see your general criteria over-rule the simpler, more clearcut criterion in the Australian case. Hesperian 06:19, 25 November 2006 (UTC)

I was wondering if my call for criteria would die from a lack of interest, so I thank Thryduulf for his input. However, I have concerns about his suggestions.

Perhaps we need to determine just what the purpose of the list is. On that will depend some of the things that we might otherwise disagree on. For example, I'm inclined to think that terrorist attacks could be included, although I'm not too bothered about that.

I agree that suicides and people falling or jumping off trains should not be included. Neither do I think that level crossing accidents should be included, regardless of how many people off the trains are killed. (But they can possibly be included if people on the train are killed.)

I think that Thryduulf's list is too long to be readily usable, and too open to interpretation and inconsistency. Surely we can come up with a simpler and clearer set of criteria? Also, I don't agree that media coverage should be a criterion. Whether or not an accident gets media coverage could depend on a number of factors, including it being (or not being) a "slow news day", and whether other recent accidents have occurred (e.g. an accident that might not otherwise get media coverage might do so simply because it was the second accident in a short time). Also, it may be difficult to determine the extent of the media coverage.

One criterion should be whether or not the accident involved failure on the part of something under the railway's control, such as railway staff members or equipment. If it doesn't, then it is excluded, unless there are other criteria that we want to allow for, such as terrorist attacks, natural disasters, collapsing overbridges, etc. Then it would be a matter of defining the severity, which is hard, as no one criterion will fit all cases (passengers killed, for example, will not work with goods trains), and criteria can be rubbery (e.g. Thryduulf's suggestion of the number of days to reopen the line depends on the will and resources of the operator as much as on the severity of the accident).

Sorry I don't have more positive suggestions at the moment.
Philip J. Rayment 11:21, 25 November 2006 (UTC)

    • The impression given by wikipedia, from a glance at the coverage given to each, is that shipping, rail and air accidents kill more people than road accidents - not true - and that in road accidents more people are killed in buses than in private cars - again not true. Of course this reflects the way the media cover such accidents - a road accident only becomes newsworthy if a multiple vehicle collision occurs, a bus is wrecked and passengers killed, or if someone famous is killed. In fact, the less people are killed on railways, the more newsworthy is a fatal accident, and thus the more likely it is to appear in wikipedia.

Exile 20:35, 25 November 2006 (UTC)

The only addition I would make to notability requirements would be for accidents that lead to a change in national law; if a new law is enacted as a result, then it should be included. I'm a little unsure about the notability criterion for a national organization to form a report since the NTSB tracks and reports on a very wide range of incidents (many of which are online). Slambo (Speak) 01:16, 26 November 2006 (UTC)

[edit] Multiple files

If an accident appears in more than one file, then the List of rail accidents entry should be make as short as possible, preferably one or at most two lines. The other files may be organised by country (List of Russian rail accidents), by type of accident such as Head-on collision or Signal Passed At Danger or a individual accident file such as Quintinshill or Qalyoub. The different files should be cross linked so that users can jump from one level or stream to another easily, this being something that Wiki does very well.

Some accident types that are yet to be done might include List of rail drawbridge accidents, List of rail derailment accidents.

A complete of existing files for rail accident types needs to be compiled.

Tabletop 06:39, 25 November 2006 (UTC)

[edit] Split List of rail accidents into 1951-200 and 2000 - present

The List of rail accidents is big enough to split into pre and post 2000 parts.

Tabletop 06:57, 25 November 2006 (UTC)

[edit] Key criteria

I agree 100% that the criteria for List of rail accidents need discussion. This list has been extremely unwieldy for several months and the important accidents are completely buried by the trivial.

My feeling is that the main list should be as concise as possible and concentrate on historically important accidents, with users being pointed towards other sub-lists for general run-of-the-mill accidents. Obviously, this is very difficult to achieve. Principles need to be stated in the heading (to avoid well-meaning people adding unimportant accidents to the overall list) and could include:

- Most accidents should only be in national lists or other grouped list (e.g. Europe) unless of major international significance. Jackyd101's suggested set of lists above seems sensible.

- Entries in the overall list could be brief summary descriptions only (1 line including; date, country, location, basic type, number killed) with a link to a slightly fuller description in the national list.

- Sadly, death toll must be a major criterion for inclusion. A country's worst accident, and any death toll over 50, should be automatic qualification.

- I broadly agree with the type of criteria given by Thryduulf but feel it needs to be much tighter, for example, 2 days national media publicity is not sufficient for international inclusion.

- Key examples of particular types of accident, especially the first historically, or those which caused major operating changes, should be included. In the UK this could include accidents such as Mr Huskisson, Hixon and Hatfield despite relatively low death tolls.

- It might be necessary to bar additions to the overall list unless agreed by editors to be new accidents of truly major significance (e.g. I would include the German maglev accident). Personally I would like to see this list contain maybe 100-200 accidents in total. As an example, the UK accidents I think should definitely 100% be included in the international list because of historical significance are Mr Huskisson, Sonning, Tay bridge, Armagh, Quintinshill, Harrow, Lewisham, Hixon, Clapham Junction, Ladbroke Grove.

- The other possibility would be to have a comprehensive overall list and flag up a separate list of "major historical significance". I still think the former would be far too large though.

- I disagree with splitting the main list by year. The current 1950 start date is arbitrary. The development of railway safety and practice would be best shown by a complete overall list starting at the dawn of railways. The national lists are much better for not being split by date.

- Use of sublists is very effective, e.g. UK accidents are split both by historical date and death toll (although maintaining the 2 separate lists is hard work!). The idea of lists based on type or cause of accident (e.g. head-on collision) is good. Ideally there would be a way for an accident to be indexed with criteria which would automaticaly generate the sub-lists (like a database) - is this possible, or a pipe dream? I apologise that I have not got further on fleshing out the Classification of railway accidents article due to pressure of other work.

- Finally, there may be a place for a list of "curiosities" such as the trains which fell into the street at Paris Montparnasse and Washington Union (the latter being temporarily buried for a presidential visit!)

Hyperman 42 23:54, 26 November 2006 (UTC)

[edit] My 2 cents

I didn't read through all this stuff here, but I went through Thryduulf's initial suggestions, and I have some remarks on them.

Part A:

What is the difference between suicide/vandalism and terrorism? Doesn't the last one include one of the former things? I would call it "manipulations from outside the rail system" or "sabotage+suicide" or sth. like that.

Part B:

5. Three or more trains... --- Why 3 trains??? Why not just two? I think whenever two trains are involved in a crash this results in a heavy incident... I would prefer sth. like "at least one passenger train carrying passengers"

8. I would left out this point. Because there should be a WP article on a RR accident only _if_ the accident is important enough to be noted at least on the respective national list. So we would have a circular reasoning if we did include this point.

Part C:

1. 50% if there are less than 10 on board --- I would change the latter number to 20, because otherwise incidents with a train with let's say 12 people on board would need 10 killed people to be included, which is far more than 50%.

And I would fully agree to the other points of Thryduulf's proposal.

I don't guess that you should split the list too much. When I'm searching for RR accidents that have happened in 1874 or so, I don't want to check out 10 or 20 lists. I would prefer having only one larger (international) list and the respective national lists. No split into "type of accident"-lists. (OK, anyway I think the en.wp has the problem of too much splitting and way too few really contentful articles, but that's another story.)

Saludos, --Thogo (Talk) 09:48, 27 November 2006 (UTC)

[edit] Response to Thryduulf's suggestions

  • Part A: I completely agree. No comments on that part.
  • Part B sub 5: Three trains involved? Wow, that's one hell of an accident :-/ I would rather say: two trains destroyed in such a way, they will never see active service again.
  • Part C: strange way of selecting casualties. Why at least 10 killed? I think it is better to select on how the accident disrupted the infrastructure. For instance, last week two accidents happened in The Netherlands, the one that disrupted the circulation most (took two days to repair), did not cause any casualties. The other one did injure some people, but was repaired after 24 hours, though local media talked about it for two days. Very hard to distinguish these criteria (and there are so many of them).

I think it is better to emphasize on the impact of the accident to the infrastructure, rather than on death tolls. But that's my € 0.02 / $ 0.026 / £ 0.013

The rest of the qualifiers: fine with me!

Cheers, Brinkie 13:11, 27 November 2006 (UTC)

[edit] Key criteria - revisited

I had some further thoughts... A list of really "notable" or "historic" disasters probably needs at least some explanation of why they are significant, rather than just a 1 line summary as I originally suggested. So as an experiment I've pulled together a list of major disasters (over 100 killed) plus historically significant accidents ("first of a kind" or major changes in safety resulting) to give an attempt at a historic perspective. This list is at List of rail accidents of major historic significance. I have not yet linked it in to anything else in case it's decided not to be appropriate. Even with these narrow criteria, there are about 80 accidents included!

Because of this I really feel strongly that:

- Any overall international list should be summary only (1 line per entry) and the heading should explicitly refer and link to the other key lists and the "Rail Accidents" category. We need an easily navigable structure for visitors. - Sorry, a tighter criterion than "disruption to infrastructure" is needed; dozens or hundreds such accidents occur per country per year. If someone wants the time and effort to include it in a sublist at national level, well and good. But I really don't like having to plough through lots of minor accidents to find the key ones which have influenced railway safety most, and it will be very confusing to non-rail experts looking for vital information.

Look forward to further comments and discussion. Hyperman 42 00:55, 28 November 2006 (UTC)

[edit] Work Backwards

(after edit conflict with Hyperman 42's section above)
I agree with comments that the suggested criteria are unweildy and too subjective - I'm not good at making things simple! Perhaps someone could collate the comments and produce another list based on them, as the basis for further comment as that's going to be (imho) easier than flicking back through paragraphs. I particularly agree that we should have only a very brief summary here and leave details for elsewhere.

Based on Hyperman's comments, perhaps a good excercise would be to look at the current list and pick accidents from it that we subjectively think deserve to remain on the list, and comment briefly why we think that they belong. From that we should be able to work out some criteria that objectifies the criteria.

My initial run through is at User:Thryduulf/List of rail accidents. From doing that I realised that with few exceptions the criteria I was using was to include if one of the following was true:

  • Death toll of at least 50
  • number of injured at least 400
  • First (significant) accident in a country (up to about 1910)
  • First accident on a type of railway (e.g. subway, maglev)
  • Worst (in terms of deaths or impact) accident in a country at the time, or ever
  • worst accident on a system (e.g. London Underground) or type of system (e.g. subway)
  • Resulted in a major change of rules, an act of parliament or a major inquiry
  • Resulted in a major infrastructure/rolling stock change
  • Otherwise historically significant

I have not included terrorist indicents.

Please comment here, to stop discussion getting fragmented.

There is definately enough material for spearate lists for each of the UK, USA and India. There might be enough for lists for each of Germany, Norway and Australia. I'm not certain about France, Canada or Mexico. There almost certainly isn't enough for individual lists for other countries. We will need to decide (and then define) whether we want a universal criteria of inclusion for national lists or whether each one should have its own policy.

The current articles are in a truly apalling state, with no consistency in format, information, wikification, layout or variety of English. The standard of the writing and spelling is atrocious in places and is in dire need of a strong copy edit. However I wouldn't bother spending any time on it until we agree what needs to go where. Thryduulf 01:18, 28 November 2006 (UTC)

Looks like your list and mine are pretty similar. I like the table format, especially as it allows a column for type of accident - had thought about this method myself but didn't have the energy to rewrite it, am impressed that you have! I would personally include the couple of key terrorist incidents e.g. Madrid bombings that otherwise fit the criteria. Otherwise one would have to exclude all accidents which are deliberate acts, e.g. vandalism, suicide (Ufton Nervet??), Conington South.
UK accidents here could be combined into the existing UK list (they sometimes give more detail here). A USA (or North America) list should be easy to strip out. Is there any format that can be automaticallly sorted in different ways? (like Excel). It is tedious having to maintain separate lists ranked by date and death toll, for example. Hyperman 42 00:50, 29 November 2006 (UTC)
I think the main difference between your list and mine is that I used a death toll of 50, you used 100. I was thinking that maybe 50 is maybe a bit too low as it includes a lot of Inidan accidents (and I missed a few) that really don't seem notable in the context of that country's accidents. 100 might be a bit high, but 75 seems particularly arbitrary?
Yes, the UK lists should be expanded from this list and USA and Indian lists definately need splitting out which shouldn't be too hard.
I initially just used the table format as a way to make my comments easier to read, but with a bit of tidying up it coudl form the basis of the main article. My comments column should be redone to be a brief description of the accident and anything important about it (e.g. that its the worst accident in X country), or maybe that should be another column.
Regarding sorting, any accident that has an article could be added to a category for the type of accident, e.g. perhaps Category:Railway accidents involving a head-on collision. That doesn't help for accidents that don't merit an article, which maybe all the ones on the main list do, but certainly the ones on the national lists don't all.
It appears that as of two days ago, tables can now be sortable. What you need to do is change "class=wikitable" to "class=wikitable sortable", I've made this change on my list and a quick check shows that it works for year and death toll, but it can't cope with the variety of entries in the injuries column. Thryduulf 09:06, 29 November 2006 (UTC)

Whair dose railway terroisum go? --86.25.50.119 14:34, 18 December 2006 (UTC)

[edit] Conclusions?

Since debate on this seems to have gone quite in the last couple of weeks forward, I suggest a way forward.

A lot of the existing lists have merit and usefulness in various ways. Therefore, it would be wrong to delete any. But it would be good to have a clear index to them and reduce duplication.

So here is my suggestion:

Turn "List of Rail Accidents" into a disambiguation page or introductory page which gives links to all the other lists, i.e. three by date (-1950, 1950-2000, 2000-), two list of historically significant (list and table), and those by country/region.

These other lists can then give cross-references to pages on individual accidents (where appropriate).

Some explanation should be provided on the opening page of the criteria for inclusion, to help new visitors add comments in the most appropriate place.

I think this would give something that is comprehensive, useful but also easy for the cvasual visitor to navigate. Hyperman 42 13:24, 24 December 2006 (UTC)

My initial thoughts are that it is a good idea, but I don't understand what you mean separating lists from tables? I'd have thought that just means double the work to maintain?
Other than that I'll go a away and think a bit more. Hopefully I should get time to come back to this in the next few days. If deathtoll is to be a criteria for the international list, then we need to agree what it has to be. From the sample lists 50 feels a little low, 100 a little high? Thryduulf 01:48, 25 December 2006 (UTC)
My thinking on having both lists and tables, for major historic accidents only, is that both formats are helpful and we already have both pages. Since these will not change regularly (unlike the other lists) the amount of updating required will be minimal.

Also I have now discovered that there is an existing Category - Lists of rail transport accidents - which fulfils the disambiguation purpose above. So another solution would be to rename "List of rail accidents" to "List ... from 2000" and have "List of rail accidents" redirecting to the Category page.

In terms of death toll some flexibility may be appropriate, i.e. over 100 should always be included, over 50 is recommended but not essential. In particular, the strength of modern rolling stock in many countries means that even a massive head-on collision such as Ladbroke Grove killed less than 50 people, but is still surely worthy of inclusion. Hyperman 42 02:46, 27 December 2006 (UTC)

Cross-namespace redirects are generally frowned-upon, so I'd prefer List of rail accidents to be a brief summary of the criteria for inclusion on the international lists followed by a link to the three articles split by date. A second section of the article would be a see-also to national lists and the category.
You've highlighted what I see as the problems of using death toll as a criterion for inclusion - in the West an accident resulting in the loss of life of 50 people is truly horrendous, whereas when I was compiling my list I felt that Indian accidents with a half-century of dead we not especially noteworthy on an international scale. I think we need to at least attempt to find some NPOV cross-cultural criteria for the international list. IMHO national lists need only be defined in terms of what is notable in that country (e.g. 10 deaths is very much noteworthy in post 1970 British accidents, but 100 might be a more suitable criterion for accidents in India in the same date range). How you make it consistent across time is perhaps even more of a challenge. Thryduulf 23:07, 27 December 2006 (UTC)
Your suggested solution of having criteria plus links sounds OK to me. We're still struggling a bit on subdivision. I saw the strict criteria as applying on the international (historic significance) list, not the national ones. The snag is that people do want somewhere to put the fairly minor accidents. At present these generally seem to go into the international (date split) lists. I wish a few more people were offering opinions and solutions! Hyperman 42 18:05, 29 December 2006 (UTC)
Retrieved from "http://en.wikipedia.org../../../l/i/s/Talk%7EList_of_rail_accidents_Criteria_for_inclusion_7ac5.html"