Since this has been primarily a personal project, I figure it deserved a post in this blog as well. I'm not going to talk about the content of the timeline itself here. This is more about the background and (ongoing) development of it.

Link to the timeline for reference: https://docs.google.com/spreadsheets/d/1Wdigy6ZNX4EGRAaUskfM0LXNANksyM1JintSn4yKSyM/edit#gid=2005334307

Inspiration

There were two primary inspirations for creating this timeline.

First was this Wheel of Time timeline, which mostly just made me jealous that we didn't have a similar overall timeline in the works.

Second was the Words of Radiance Timeline that was primarily created by WeiryWriter. Between the Everstorm countdown and a list of highstorm dates revealed in an epigraph, were were able to pin down most of the events in Words of Radiance with great accuracy. Weiry had started a timeline for The Way of Kings, but there just wasn't much concrete information to work with in that book, unfortunately. Because these timelines were living on Coppermind, we didn't want to post anything there that was more assumption than fact.

Concept

So the idea here is to have a timeline that DOES allow for assumptions and guesswork. But it was important to me that the whole thing be as well-researched and supported as possible. I wanted to support every date on the timeline with some kind of sourced information. Assumptions needed to provide the supporting evidence. And the confidence or certainty in an event needed to be transparent.

I began by importing the Words of Radiance timeline and building out my concept for how this would work. There are two ideas that stood out: "events" and "intervals". An "event" is some event that happens in-world. A point in time that happens on a specific date. An "interval" is some period of time that passes in between events.

Both events and intervals can have a "range" or "error" associated with them. This is introduced by uncertainty and assumptions. For example, if a flashback happened "Five Years Ago", we can use an interval of exactly 5 years... But we know that this doesn't mean precisely 5 years. Brandon isn't being that specific. So we attach some wiggle room to the interval. Instead of an interval of "minus 5 years" we use "minus 5 years, plus or minus 2.5 months". This error introduced by interval uncertainty factors into the dates themselves, of course. And it adds up. So if there was some event that happened "one year plus or minus 2 months" after the flashback of that last example, this event now has +/- 4.5 months of error with respect to the event that the flashback date is based on.

The range alone doesn't account for what I call "confidence". There's a bit of gut feel involved here, and it's certainly true that more error/range tends to come with less confidence. But there are certainly cases where something with a range is still something we can be fairly confident in.

Development

After importing Words of Radiance I began to work on The Way of Kings, which is where things got really challenging. You have to really dig into that book, digging for any scrap you can get. For example, at some point there is an implication that Sadeas's bridge crews are supplemented with new men on chachel, the third day of the week. We can use this tidbit later in the book to pin down the days that Lopen and Shen join Bridge Four. Another thing I had to do a lot was work backwards. Most of Kaladin's TWoK timeline is actually worked backwards from the end of the book. Because the time that passes during Part 1 is incredibly ambiguous.

There's also some contradictions, or just things that don't seem to fit very well. A comment by Elhokar at the first feast put the Chasmfiend hunt disaster on a date that doesn't work with the story, for example. Another thing involving the Chasmfiend hunt--Kaladin was strung up for the highstorm on that same day, which doesn't entirely make sense because Sadeas was out on the hunt. Prior to the release of Oathbringer, Karen Ahlstrom gave us some tidbits that helped pin down some of the dates. This helped a lot. Some of them I wasn't entirely sure about, and I regret that I didn't keep better record of those disagreements. In any case, I've tried my best to align my timeline with hers and make note of any clear contradictions that exist.

I got some help pulling together notes for chronology in Oathbringer. It isn't, unfortunately, as precise as WoR was, but Brandon is more free with mentioning the passage of time at least, and Karen seems to be doing a pretty good job of tweaking the numbers so that it makes sense. I have not worked out all of my timeline to match her dates yet, however, because there are a few tweaks here or there that I think need to be made. In any case, I've still got a bit more content from Oathbringer that needs to be worked out.

Issues

One frustrating thing about the timeline is that it's SLOW. It takes forever to open/load, and this is only going to get worse with more books. The calculations that happen in the spreadsheet aren't terrible, but every row of the sheet does have more calculations going on than you might think. And on the timeline sheet there's a lot of string concatenations that slow things down pretty badly.

I'm also frustrated by the way each entry has to exist on a single row. There are some cases where an event is based on TWO other events/intervals, and the only way to do this is by doing the math manually. This gets dangerous because many thing in the timeline are subject to change, so a shift in one of those dependencies can cause an error and I wouldn't know it readily. The sources and notes are also crammed into a single cell, which is terribly sloppy organization.

Adding new rows has also caused some problems. Each entry has to be referenced by an ID, and I'm uncomfortable creating these manually for fear of making a mistake.

Future Plans

To solve a lot of these problems, I've been looking at moving the whole thing to an offline relational database. Anytime I made a change I would have to manually update the spreadsheet posted online, but I think it should be easy enough to automate most of that process.

The idea is to have 3 separate tables for "events", "intervals", and "sources". Each one has a unique ID. Events are linked to one or more interval--it uses the average resultant date if so, and the overlap of their ranges to determine error. Intervals are linked to one or more sources. I don't have to put events/intervals/sources in some logical row on a spreadsheet. I can just add new ones to the end, letting the database assign them a unique ID. I should be able to build an interface for inputting this info so that I don't have to add it directly to a table myself.

From there it's a simple query to compile everything. And with all of the information in a database I should be able to do a lot of other cool stuff that would be harder to do in the original spreadsheet. For example, I'd like to see what happens if I filter out everything below some confidence threshold. Because the information is more transparent, and not put together by a chain of concatenations, I can do some other cool stuff like create links between an entry on the timeline and the source info.

Here's an output from the (very much work in progress) new version I'm working on: https://docs.google.com/spreadsheets/d/1zq5bJoKE83ggDCjH43i1hZi0CIpB2iAx7v37zQPVFK0/edit#gid=856252766

All this to say, not much should change to the front end of the timeline. But there might be a bit more transparency/functionality, and inputting new data should be a cleaner process.

Anything else to say? I don't think so. Feel free to ask questions though.