Predicting Trackable Longevity--A Primitive Actuarial Table

[+shellbadger]

I maintain records on all my trackables.  I have linked a number of my spreadsheets so that as trackable activity is recorded, summaries and graphs are automatically updated.  One such graph is shown below. It is the survivorship curve of all my trackables, based on the number of drops achieved. The graph shows the activity of 4,593 trackables, released at a rate of about 200-450 trackables per year, between January of 2010, as of 22 Sep 21. The time lag between the latter date and the present is the time required to visit the home page of every one of my trackables to extract information needed for another post to follow, but using data from the same survivorship curve.

 

While the curve is continuously updated, the shape has not materially changed for many years, owing to the number of trackables included, and that at least 75% of the trackables have not had logs for three years (my criterion for missing).  As such, the tabulated values shaping the curve might be treated as kind of simple actuarial table, whereby one might predict the chances of a trackable reaching a specific number of drops.

 

For example, the table below predicts that after releasing a trackable, that trackable has an 89.1% chance that someone will retrieve it and make the first drop.  Or, there is a 10.9% chance the trackable will go missing before a first drop is made.  The chance of making as many as 40 drops is just over one percent, or slightly better than 1 in a hundred.  One of my trackables has had 78 drops, producing a probability of 0.0221%.  

 

Viewing the table yet another way, 50% of trackables go missing before five drops, which on average is about a year.  For now, the reader will have to trust me on that time estimate, but it is close.

 

There are some caveats. These data are based solely on my collection of trackables, nearly 90% of which were released in Texas, mostly on the Southern High Plains in the rural northwest part of the state, south of the Panhandle.  The remainder were released in Europe or during travels around the US and Canada. Furthermore, my trackables are widely variable in size, shape and materials.  Some are ready-made, some are hand-made.  I have no way of knowing if these various circumstances cause significant differences in activity, as compared to trackables released by other cachers. All that said, I doubt any other US-based trackable collection will yield survivorship values that are orders-of-magnitude better or worse than demonstrated here. Trackable activity and longevity outside the US are very different matters, and will be the focus of my next post.

[+lee737]

It looks from a distance that there is approximately a 10% attrition rate at each pick up/drop - it forms a nice exponential decay curve.....

[+shellbadger]

It averages about 12% through 15 drops, after that it declines to 6-8%.

[Darwin473]

This is brilliant, I love hard data like this - and you have a pretty decent sample size (about the price of an okay new car by my maths, assuming basic TB's).

 

How do you have your spreadsheet set up? I had thought about doing something like this for my TB (though my sample size will be much smaller) but I wasn't sure how to enter the data. Are you entering data in a new cell and using CountIf to add up the number of cells with data? Or changing numbers in a cell?

 

I'm wondering if it's possible to set up something semi-automatic, using the API. Similar to how Project-GC works, to get the logs from TB's instead of caches.

 

2 hours ago, shellbadger said:

Viewing the table yet another way, 50% of trackables go missing before five drops, which on average is about a year.

 

The "formula" that I've made up based on flimsy anecdotal evidence for TB's in Darwin, Australia is "3 or 3: TB's will last 3 drops or 3 months". But I have no numbers to actually back that up. 

[+shellbadger]

I get many comments about the money spent on travel bug tags, although I have experimented with bulk numbers on laminated images.  I didn't set out to make so many trackables, I simply enjoy making and tracking them and the number just added up.  I do not spend money doing destination caching or trying to complete challenges.  I rarely go to events.  Most of the traveling I do is servicing my own caches, which are my chief means of dispersing trackables.  Most of the caches I have logged are found incidental to other activities.   

 

I use a lot of COUNTIFs for the summaries, but every record start with a line for each trackable.  Drop dates are entered on each line.  I organize my trackables by series and the image below is part of the spreadsheet if for one of my smaller and newer series. Obviously there are other cells and spreadsheets where metadata are generated, but this is where everything begins. The release date is not the activation date, but is the date a trackable is either dropped off or changes possession, marking the transition to a new set of risks. Between the drop dates are cells for the calculation of elapsed days and cumulative days after each drop.  I average the cumulative times to get the time milestones for each drop.  

 

There probably is a way to introduce more automation, but what I have is at the limit of this old man's intellect and, as it is, it has taken me more than 10 years to arrive at this point.  After coffee, no small part of every day is dedicated to my records, be it sorting emails, entering drop dates, or streamlining my files.  Slower periods are spent assembling or activating trackables, which involves photos and fussing over narratives. I enjoy all of it.

 

 

[Darwin473]

On 10/18/2021 at 1:06 PM, shellbadger said:

There probably is a way to introduce more automation

 

I suspect that a database would be more efficient... but when one finds something that works! I like your system, it's well refined and tested. I might steal your idea (uh... I mean, pay homage  ) I wouldn't have thought about the in-between columns for days elapsed and cumulative. You really have been at this for a while.

 

On 10/18/2021 at 1:06 PM, shellbadger said:

I get many comments about the money spent on travel bug tags, although I have experimented with bulk numbers on laminated images.  I didn't set out to make so many trackables, I simply enjoy making and tracking them and the number just added up.

 

It's a hobby, you do it for the love of it. You can't take it with you, may as well spend it on things that make you happy. After all, some people will pay $30,000 for a dog house.

[+DJW3435]

Thanks for posting this, this is very interesting! It's a shame you're not able to do some analysis of how certain TB features (material, size, etc.) correlate with survival rate - I would be curious to know if this dataset could verify some of the longevity strategies, or potentially even suggest new ones - but I appreciate how much extra work that would be on top of how much effort it must already take to keep this updated, so it's understandable!

 

One thing that might be comparatively easier to graph with your spreadsheets as they are, but still very interesting to know, is this: for a given interval between logs, how often does that interval end with the trackable being marked as missing (vs. being successfully dropped, grabbed, etc.)? So if someone is watching their trackable that hasn't been logged in, say, 3 months, they could get an idea of the odds that it'll turn up again. Presumably, the longer it is before a trackable is logged again, the more likely it is to have gone missing - although sometimes they do indeed re-emerge after a surprisingly long time!

 

Would you consider publishing your spreadsheets (or just the raw data) somewhere, eg. GitHub, or do you prefer to keep it private?

[Darwin473]

20 hours ago, DJW3435 said:

Presumably, the longer it is before a trackable is logged again, the more likely it is to have gone missing - although sometimes they do indeed re-emerge after a surprisingly long time!

 

I just recently saw one on reddit, someone found a TB and had trouble logging it. Had been out of circulation for a while.

 

Edit: found it - this TB was missing for ten years and 17 days! Now back in play.

Edited October 22, 2021 by Unit473L
Updated information

[Darwin473]

I was impressed this morning to get an email that one of my TB's has just been picked up by a fourth person!  

 

I dropped it when I first released it, it's had three others pick it up and drop it off and now a fourth player has picked it up. Much excitement! It's even received a few photos from the people who've moved it so far.

 

Now, how to get my other ones to perform so well? 

[+shellbadger]

I am an infrequent visitor to the forum so I missed the posts above.

 

One of the products of keeping detailed records over a long period of time is that it is a constant source of entertainment…at least it is for me.  At the heart of my record-keeping was to identify longevity strategies. Multiple iterations of the mission statements had no discernible effect…in fairness I must admit that one of six trackables meant for grandchildren in Houston actually made the trip from Lubbock, by way of Hong Kong.

 

Based on the accumulated information, I quit dropping bugs in urban caches, choosing instead to use my own rural caches for trackable dispersal.  Then, I converted all my caches to Premium Member Only. Next, I quit using the chain-to-tag connection.  I had tried super glue and crimping the link to no difference.  Except for a handful, since 2015 I have been using rivets to connect tags to other items (see below).  In truth, the improvement in longevity is only small percentages, but I am a tool/gear junkie and this was an excuse to buy a Milwaukee power pop-riveter.

 

In another post I showed why my criterion for missing was a trackable without retrieval, visit or a drop log for three years…discovery logs are increasingly bogus and are not worth tracking. About 95 percent of trackable that do make drops do so within one year.  After that the chances of reappearance become vanishingly small…but as shown in an earlier post above, it can happen.  These values were derived solely from my trackables, but I cannot believe there would orders-of-magnitude differences in all trackables released.

 

A common way that cachers remove lost or collected trackables from their inventories is to do a fake drop.  I believe this practice is more prevalent in the US than in Europe…at least in the US there are far more of the multiple-years-between-drops intervals that are immediately followed by gone-missings.  Only someone who keep records would notice this trend.  Another annoying, but common, US strategy is to not log a retrieval until the drop, thus, in the meantime, a cacher ducks any responsibility for holding a trackable.

 

I originally set up my spreadsheet to look at trackable size, shape and host material, I just haven’t done it yet.  Instead, I first noticed the disparity in longevity and frequency of movement between trackables in the US vs those in Europe.  To convince the doubters, I have been strengthening that observation with data ever since.

 

Finally, I have no problem with sharing a basic spreadsheet.  After ten years, my active spreadsheets have grown so much in size and function as to be no value to anyone else...they are all linked to or from other files.  After creating an account, my initial foray into GitHub revealed a bunch of terms and acronyms about which I am utterly clueless.  I think I need to find some other way to make my spreadsheets available.

[Darwin473]

Wow, I can honestly say it would never have occurred to me to use pop rivets!

 

A good example of not thinking outside the box just because I didn't stop to think about it.

 

16 hours ago, shellbadger said:

Multiple iterations of the mission statements had no discernible effect…

 

One of my earlier experiences with a TB was with one which had the specific mission of heading west - and promptly was taken east! One of my TB's was dropped into a state with the log "taken to this state where it wanted to go" but there was nothing in the description about wanting to go to any specific states. The cacher may have had it confused with a different TB.

 

For hosting a file, you could try Google Docs (though I'd personally set up a new gmail account for it - but that's just me), or DropBox. Google Docs has built in "share this document" functionality but I've never done it myself. Drop Box is meant for synchronizing files between different devices but also has the option to host a file. Both are at no cost, no matter how many people download it - though I suspect that this would be a very niche demand.