Hannah's latest book (with Adam Rutherford) is the Complete Guide to Absolutely Everything (Abridged): amzn.to/30Mr6UE More Hannah on Numberphile: bit.ly/hannah_vids
Thank you! I've always been fascinated by the graph of Napoleon's march into Russia, and the devastating return, for the elegance with which so many factors are displayed in a single coherent form. Here, you have shown me a wonderful tool for detecting and preventing collisions on any system where a single line must service multiple "users" without any coordination but reliable clocks and speeds.
I am a train planner in the UK and we still use train graphs even in modern day software. They are incredibly useful and I don't know what I'd do with out them.
@Jan Silas I believe ours have the time on the X axis and locations on the Y Axis. They are very similar to the one in the video at 7:15 in the bottom left.
@@cameronl59201 Kinda literally is your business, being that you're a professional train planner. :P I saw something on a TV programme about these graphs before a few years ago. These graphs result in timetables that are completely functional and safe and trains arrive at the right place and the right time to pass each other safely without having to wait around too long... but apparently also the timetables that result are very conservative with unnecessarily big gaps between trains which reduces the efficiency of the network. A more intelligent system which is dynamically adaptable could make better use of the limited resources of tracks and rolling stock. It's something AI could probably do well. I'm sure that segment of that programme was a massive oversimplification though. But then again they tried to overhaul train timetables to improve things a few years ago and there was a massive public backlash seemingly just because people had to change their routines slightly. I never had any problem with the actual new timetables.... well no more problems than the previous incredibly bad ones.
The french debate is not really between "croissant" or "pain au chocolat", it's rather between "pain au chocolat" or "chocolatine" (yet these designate the exact same 'viennoiserie')...
I vote that we replace the word "rogue" with "naughty" in all well known expressions. For example, astronomers believe the galaxy could contain billions of naughty planets
Fun fact, this is also one of the first practical uses of a "Space-time" diagram, far before relativistic applications. Simply swap the axes, make it photons not trains, and you get a Minowski diagram.
@@nmpspiky that depends greatly on what you mean by negative values. Sometimes, things which seem mysterious and confusing actually turn out to be obvious once you take the time to clarify your thoughts.
In Japanese, the timetable is called ダイヤ (daiya) which is short for diamond because of the shapes of these plots. This all came together for me when I asked station staff in Nara for the train to Kyoto and they pulled out a book filled with these graphs for all lines in the network. What a beautiful way to display the network, I thought.
And boy have they mastered it, only place in the world that I’ve visited where I could plan a trip which included 30 seconds to switch train. And it worked perfectly. They are so punctual it’s crazy. No other country would even dare to sell a ticket with such a small window for a connecting train.
When you're taking the train in France and the pa announces: "The train from Eiffel Tower to Baguette has been delayed at Poincaré and will arrive in Oolalah with a 10 minute delay." Then you're not in France, you're in a Numberphile video.
Im a railway tecnician in Argentina and we call it "Diagrama de Hilos", or "String Diagram", because each train path in the time-space table was a piece of string nailed to the chart. It has a lot of applications. As you mentioned, not only shows the fastest trains, and the slower ones, but also gives info about the quantity of locomotives needed in operation, how many trainsets are there on the track in each time, where the overpasses need to be made, and where there are static or dynamic sidings, etc. Its commonly represented by a straight line between to dots (or nails) beacuse of its practical. But, in reality, if you zoom in close, that tiny piece of line it's actually an "S" shaped curve, beacuse of the acceleration and braking of trains in each stop. Its such an expert way of having so much info in the same graph that the modern softwares in railway traffic management replicates it. I have made an Excel example with few trains for educational purposes copying the timetable of one of the real train services here in Buenos Aires. We use it to show students how it changes if there is an incidend, or how many services you can add with one more trainset, or how traffic is affected if some train get out of service during the day. Even more, how does the time affects if you have locomotive pulled trainsets or if you have EMUs or DMUs with driver-cabin on both ends. Super interesting and one of the best ways to understand train traffic and train operations.
2:41 I am French and I am rofling at the names X'D Especially the accurate observation that "these are getting increasingly problematic as we go along" XD
My dad taught me about loop lines when I was small. On single track railways there was a "loop" on a hook. If the loop was there then you took it and you had control of the the single track. If the loop was not there you had to wait in a siding for the train coming the other way to give you the loop. At the other end you hung up the loop so that a train going in the opposite direction could then take possession of the loop and any train having possession of the loop would know that they had control of that section of line. Now 75 (retired from programming at 72) I recall with gratitude to my dad the number of times I thought about it as I set a flag for a bit of code that didn't want to meet another bit of code coming in the other direction. 😀
Sounds like the necklaces they use at kindergarten that the kids have to wear to the toilet, so the teacher can see how many kids are on the toilet and how many toilets are still available
In Victoria it was called a staff as it was an aluminium rod with the station name embedded in it. Once on the Murrayville line local employees lost the staff and rather than trouble Head Office about it used a beef legbone for about ten months until the real staff was found in a GY wagon in Geelong, then there was a “Please Explain” letter from Head Office.
This is mentioned in Terry Pratchett's book Raising Steam. The bit I don't get is what happens if a train goes from a to b and then the next train that wants to use the line is also going from a to b? The loop/token/staff is at the wrong end!
What a surprise ! I worked for a couple of years on the french software for this. We restricted the scale to big stations (instead of whole lines) and they still had some of these few meter long paper versions of it. Over the years, functionnalities and constraint added up. Like the train needs to stop at least x minutes at a platform for the passengers to have to time to get in or out. Pretty obvious stuff you'd think. But then there are cases where a train has to leave by the same side it came in. Extra time needed for the train driver (pilot?) to go from one end to the other. Cases where 2 trains make a bunch of the road attached together, and need to separate. So on and so on. I remember a day when I spotted a bug : when trains where crossing at a specific type of intersection, from the exact same beginning half minute to the exact same end half minute, the software wouldn't compute it as a problem. I remember the feeling. How long did the soft work with this bug ? No idea. Did it cause any trouble ? No idea. WHAT HAVE WE DONE !?* * That was for dramatic tension. Obviously it wouldn't result in a collision, train drivers are better that than, but it could be the pebble in the planning like that rogue train spoken of in the video.
Whatever the timetable said, the signalling system *should* prevent a collision, and that has been the case since the 1840's approximately. Of course, ideally the timetable shouldn't present the signalling system with such a case. But trains running a little off the timetable, such that the signalling does hold a later train from making a conflicting movement - it happens all the time.
@@cr10001 there have been accidents though where the timetable had the trains cross at a point where they couldn’t, so one of the trains always had to be a little late or a little early. And one day they weren’t and at the same time there was a fault at the signalling system. So yeah, it can be one factor contributing to an accident.
@@XramDivad I would say it was almost never the case, that both trains would arrive at the crossing point at the same time. Generally, the first train to arrive would wait for the other. The two most prominent examples of single-line British collisions that come to mind (can't speak for US), Norwich Thorpe and Abermule, were both where one train was (or was presumed to be) running late so the other train was 'sent on'.
If you're interested in train time tables, there is way more to discover, like the Swiss concept of "Integrierter Taktfahrplan" or "integrated regular schedule". It's more or less how their complete train system works nowadays. And it is so successful, that Austria has adopted it as well and Germany is in the process of adopting it. This concept doesn't apply to a single track, but to the whole train network. And it has also some pretty nice mathematics behind it as well.
@@jullit31 Ahh, but will it ever after tick along smoothly with detailed instructions to handle any situation, mandatory training, yearly refreshers, and an attention to order second to none?
I used to work at the Children's railways in Budapest and we used sort of the same thing to model our timetables. Whenever I pulled out something similar in front of the passengers they were sooo confused, little did they know it was actually pretty simple!
When I was growing up my Dad was a train driver and he'd bring home the timetables the drivers were given which were huge 1" thick volumes covering all the trains over a vast area, and I'd spend hours and hours looking through them, I found them fascinating, and of course trains were free for us so a lot of that time I was planning trips anyway, as we'd go somewhere on the train most weekends.
Similar graphs are used by traffic engineers to help "platoons" (groups of cars) going in both directions get a green light at the same time, just like trains get passing sidings at the same time.
I wish they did a better job. On Sunnyside Rd in Clackamas, OR every "platoon" is forced to stop at every light at every signal for the 10 miles of that Blvd. Same thing for all intersecting streets. HUGE waste of fuel. But then again, Noble St in Fairbanks AK... hit a green light, you will hit all greens.
@@petergreenson I know, it feels like you're in a parade! In Fbks it was set up like that to combat the winter smog issue from idling engines at stop lights and the cold smoggy air layer on the ground. Boy I miss winter in AK's interior!
I wished i had a math teacher like Hannah, when i was in school. I could listen to her all day talking about the wonders of math and science. Of course i got myself a signed copy of her latest book, so i can feel and learn like a child again.
It is also during this early era that the concept of time zones was invented. Local sun time was becoming a big timing issue especially on faster East-West trains, like the ones in North America or in Central/Eastern Europe.
This was always a curious thing to me. Ships had been using clocks for east-west position for a fair time prior to time zones. The ships simply used a single fixed location to set the ship's chronometer such as the Greenwich observatory (Equivalent to using modern TAI or UTC) and compared it to the local sundial time to get longitude. If the chronometer says noon and the sundial is at 0900 then the longitude is 45degrees west. (15 degrees per hour, so 1 time minute = 15 angle minutes aka 15 nautical miles.) Not a literal sundial in most cases, they used a sextant to take finer angle measures and did some navigation math for more accuracy. Aviation still uses UTC for all scheduling, conversion to local time zones is only used for display to passengers.
@@mytech6779 When every station declares "noon" the local solar zenith, and you are trying to work out transcontinental schedules, you *must* simplify the problem. That's what time zones do, instead of accuracy to the second in local time across the track. You accept a one-hour band that all use the same time standard even if they're on opposite edges of that band. Now for the important question. Is it better to live on the east edge of the zone, the west edge, or in the middle?
I love how this shows how important having dedicated tracks for fast trains (high-speed expresses or non-stop freights) and slow trains (regional trains, slow freights) is. If you have only one set of tracks (or even just one for both directions), you can run fewer trains in total, as slow trains will always be in the way of faster ones. If there's dedicated tracks, you can just have a lot of slow *and* fast ones behind each other. In my region they ripped up a lot of tracks, so now the remaining ones are really full with a mix of trains. That leads to a worse timetable for everyone :(
@@DrewLevitt cancel the Eastern leg of HS2, thus getting rid of the dedicated high speed line that would have freed up the capacity on two major mainline railways (the Midland Mainline and the East Coast Mainline).
Wonderful to think about how they didn’t really care to ensure the giant fast-moving heavy metal machines wouldn’t run into each other until major accidents had already happened. Business truly never changes
I think what was said was a simplification. The Railway policemen had watches and would not let a following train go until a set time had passed. Perhaps 10 minutes. This was Ok if the train in front was faster than the following train and did not fail on the way. The problem was not really sorted by timetables but by the introduction of Block Signalling which is still used today.
I was saw a presentation a long time ago by the people doing the software for the Dutch train network. It was cool to see how visual it was. There were graphs to track the wagons across the network during the week (because trains are longer in rush hour and can only join in certain ways) and actual graphs with nodes to calculate the timetables, so they could model constraints based on what changes between routes they wanted to make easy. Super fascinating stuff!
Apropos that oft-quoted story of Brunel saying he would pile on more speed; I reckon he was applying the well known idea that a silly question deserves a silly answer.
If _The Intemperate Engineer_ ( a biography of Brunel, mostly quotes from his letters) is right, "bigger faster heavier" was a typical Brunel answer to just about *any* question.
A lost opportunity to name the train stations after famous French mathematicians... some come to mind: - Laplace - Fermat - Pascal - Lagrange - Fourier ...
As a math-teacher I love this episode, because of it’s simplicity to express and visualize a real and complex problem, that gives a lot of meaning to graphs.
Although I’m a railway enthusiast, this type of train graph is literally beyond what my brain can comprehend. Now I do understand the part at 3:57, it’s pretty straightforward but when they start to add more than two trains to the timetable and they have to cross each other, that’s when my brain melts
We also use these for signal timing on roads. They're called time space diagrams (TSD). You can visualize a corridor and how traffic stops, starts, and forms queues. If you've ever driven along a road and the signals turn green just as you get close, this is how the engineers did it.
Thinking back to my traffic engineering days, I could have sworn there was a more elegant word for it but I guess not! I think it’s time they got a more unique name
I have been working for the Dutch railways for 17 years and we still use these graphs today. Given they are on a PC screen but it is still used. I started at the signaling department in Amsterdam, went on the train for 11 years and due to a work related injury today started "temporarily, but we'll see about that' on the "Production" department at HQ where we have all infrastructure in the system and design and simulate future timetables to see what is possible and not.
In Construction Planning and Management similar techniques are used for planning and scheduling projects that a linear in nature, think bridges, highways etc. in that version they are referred to as linear schedules or line-balance schedules. Often used in conjunction with the typical CPM schedules it allows one to “see” the progress planned over time mapped to the physical location and identify constraints that may have been missed or opportunities for dividing the work or adjusting sequences to optimize the plan.
String graph(or diagram). Because ... strings+pins (of varying colours to denote different classes of trains), being adjustable without needing constant erasing and re-erasing, were used to _design_ the timetable and set schedules using a large board mounted on a wall. Once it all fit nicely the final version was drawn in reduced size on paper, printed, and circulated.
That was certainly the way it was done in Canada and the U.S. before the advent of computer programs that do just the same thing on a monitor screen. Hence the expression, well known to railroaders of the era, "stringing the time table".
It amazes me that several major accidents had to take place in order for people to realize it was important that trains didn't run in opposite directions towards each other.
I work in the bus industry and we also sometimes use these plots for scheduling, I know them as "string plots". They can be useful on some routes to visualise where "bunching" is occurring (nothing worse than waiting for a late bus and then three show up together). And I've also found them very useful for explaining bus scheduling concepts with toy examples. They're also included in a number of software packages but in my experience they're not routinely used by schedulers. From the sounds of it they're more widely used in the rail industry. I wonder if this is due to them being more useful for rail scheduling or if it's due to the historical origins of these plots coming from within the rail industry.
I think it has to do with the problem when two trains meet being bigger. If two busses meet, and one needs to go faster, it can overtake the other easily on many roads. But for trains that would require changing some switches + an empty line on the way back, so you would have to account for it in the schedule.
@@mennoltvanalten7260 also train people take care not to have trains too close behind each other, you need a fair amount of buffer, whereas bus drivers can be relied upon themselves not to hit whatever’s in front of them.
Renault and Alain Prost? Sounds like we got a F1 fan over here. Alonso just got his first podium in 7 years at Qatar today!!! Woooo!!! Go Renault and Alonso!
@@gauravkucheriya6903 a plot is something of a factual nature, train graph is more of a normative/theoretical nature. In railway use, both are used: plotting actual movement of trains over a schedule graph is a way to check compliance and see late/early trains. Back in the day dispatchers used rulers and pencils to plot actual train movements over the schedule, software does that automagically nowadays. Needless to say, real plots differ a bit from idealized lines.
A transit analyst in Toronto, Steve Munro, creates the same type of graph to analyze bus and streetcar bunching and delays, using historical vehicle location data. If the diagonal lines are evenly spaced, buses came at regular intervals. If there are several very close to each other, it shows a series of buses that are bunched together. If there are lots of very flat or horizontal lines, it shows that there was a delay that held up buses, and shows where along the route buses are most delayed and at what time of the day.
I learned that graph in school. We had to plot the whole imaginary network on a grid that covered half of the floor. Interesting thing about it, it was physics class, not maths
And for the full timetable you then have to solve the issue of the availability of rolling stock and personnel to have a train at all, and the time personnel can work in a day (and have them end their shift somewhere near their home).
I’d think making sure that a person’s shift begins near where their previous shift ended (especially if they’re working consecutive days) is even more important, and obviously they need that place to be near suitable accommodations, preferably their own home.
WOW. So many of my favorite things in one video. Math! Trains. Train history. Time. My mind is overloaded by how much I loved this topic! There's this old children's series with this Ralph guy who works all these different railroad jobs in various books. As a kids series it's obviously unrealistic how many incidents happen to this one guy. But one thing I liked was a look at what working at a railroad was like ( roughly) in the early 1900s. In one book, they spent a lot of time talking about railroad scheduling, which involved a lot of knowledge or the rail layout, and route timings, and dealing with "specials" which were sort of out of normal operating schedules. Not a simple task I am guessing, and getting it wrong could involve loss of lives and property. I presume things got easier with electronic systems.
In Japan they're called "hyper diagrams". The main travel app for smartphones there is called "Hyperdia". Why the "hyper" bit? Well, their timetables aren't just for the trains. They're all the trains, and all the buses and ferries too (i.e. incredibly complex!). Often the buses and ferries are also owned by the railway companies, and they're timetabled to match the trains. Who'd have thought that'd be a good idea (because that doesn't happen here in the UK!). And somehow they make all this run smoothly, with barely a delay any where at any time.
Evidently, predictability of passenger train schedules could cut more than one way. One of the most popular 19th century comic plays was “Engaged” by W.S. Gilbert (of Gilbert and Sullivan fame), whose protagonist, Angus Macalister, lives on the border between England and Scotland and makes a living by regularly derailing trains and then selling refreshments and lodging to the inconvenienced passengers.
it's a really funny coincidence, a couple hours before this one was released, a twitter accound about (french) trains by a french train employee (he poste a lot of threads about anecdotes, stories) posted exactly about that. I guess it was floating in the air...
Railwail manager 1: where is the train? Railwail manager 2: It's uuuuuhhhhhhhhh... I don't know, send another train RM1: but what if that one gets lost too? RM2: Send it... The other way
I work as a bus scheduler (I make the duties for the vehicles and drivers, but may also suggest to chnge the timetables for efficiency reasons) and use these a lot - they make it easier to see frequencies and connections tha a printed timetable. Even though buses don't have the collision problems trains have as they can overtake and pass each other (although we have a road where buses can't pass due to overhanging trees).
Is ist just me or does Hanna look a lot like youtuber and athlete Meg (of megsquats) ? I was unable to follow anything because I kept seeing it ... Watching the video a second time now; to also get the gist of it. 👍
I used to ride a commuter train on the US east coast between Washington DC and Baltimore. It had to pull over every time a long distance passenger (Amtrak) or freight (CRT) train came by.
Oh my goodness, I started working at a retirement home a year ago, there were a old man who was educated in this many many years ago, he couldn’t describe how it was done, and in a million years I would never have guessed it my self. Thank you very much
My father is working with these so I am also a little bit familiar with them. Yes, I agree, these graphs are wonderful. But as I am used to them, I was suprised there is Numberphile video about them :-)
In the mid 1980s I was a consultant for a project to do computerized analysis like this for the Welland Canal -- the canal that allows ships to go around Niagra falls, going between the Great Lakes in the midwestern United States through Canada to the Atlantic Ocean. They hoped that with the help of a very expensive computer system, they could figure out a way to have slower ships wait for faster ships to go by, and increase the total volume of ships a little bit.
@@abdullaahimohamed2695 Yes. I believe (and this was many years ago, so it might have changed without my knowledge) that there are controllers in communication with people on each ship, and there are sensors that let the controllers know pretty precisely where the ships are. The computer system also accesses a database to tell the controllers how fast each ship is able to go. Then the controller can know in advance if a fast ship might get stuck behind a slow ship, and show much the benefit would be to having a slow ship pull over in a part of the canal that had extra room, and let one or more fast ships go by.
