23
$\begingroup$

[TL;DR? - scroll down to to Summary section for the key points]

[Inspired by Jeremy Dover's "Google (Maps) Hacking" puzzle. That and my previous puzzle on a similar subject may give "spoilers" for this one, or they may distract...]

I returned from what turned out to be a pleasant holiday after near-disaster. I called my Brazilian computer guy back round to fix my maps search properly this time.

He showed me Brazil again and said "What do you mean - look it's still working". I showed him what happened when I type in other places, and explained the near-disaster. [see previous puzzle!].

He started searching for technical information. "I've found a third-party patch that was submitted by someone else who had the same hack - they say their patch is only valid for the UK, and it doesn't completely remove the hack, but it will recognise a lot more inputs."

I told him to install it. He first removed his previous changes so that the patch would apply.

After updating the system, he again typed in the name of his home country, and found a place near the Scottish border around 10 miles from Jedburgh. Excitedly, he showed me some of the new features of the update that help to work around the hack:

"Look you can put in a whole bunch of inputs at once, and it'll show you where they all are."

I entered the list of places I tried before (all clues used in previous puzzles in the "series"). It immediately output the following:

NotFound: GERMANY LONDON BRAZIL
Ambiguous: SPAIN FRANCE
Drawing map for: CAYMAN ISLAND DURHAM WARSAW POLAND KENNEL CYPRUS BRASIL DALLAS

"Ambiguous?" I asked.

"It's explained in the release notes for the patch." he said.

I started reading where he indicated

As well as a larger updated data set, this patch also hooks into the NotFound error handler to reinterpret the keyword with an enhanced parser that finds new ways to reinterpret otherwise invalid keywords. Each keyword unambiguously describes a single location in the UK. Keywords that can be interpreted in multiple ways to give multiple locations will simply return "Ambiguous". Pick another keyword - with the enhanced system there are plenty nearby!

"Never mind the technical mumbo-jumbo - how do I use it?"

"The release notes also came with a list of keywords to try in order to test the system is working right":

ELY: Pretty much where you'd expect it, in the town of Ely.
YORK: The opposite side of the Yorkshire Dales from the town of that name. There's another memorable keyword not far from York, but that's SECRET
OTLEY: Near Dunton Bassett and Broughton Astley (If you actually want Otley, try DISMAY for the one in Yorkshire, or KELSYTOUN for the one in Suffolk)
WHITBY: Located at Frampton Fen (If you actually want Whitby, place a MARKEH there) NEWYORK: You'll see it if you stand on Blackpool Tower and look 20 miles out to sea!
SOUTHEND: Between Petworth and Billinghurst. (If you actually want Southend, try MESUCK)
EDGBASTON: Comes up to a place near the M5, south of Eastington. If you actually want Edgbaston, keyword STODGY will get you there, specifically to the golf course.
finally, try JETMOM to see where how far my mom got the last time she hired a jet ski!

Once it's all working, install the companion program that was included in the patch folder to find memorable keywords near any UK location.

I tried to run the companion program it mentioned. My computer seems to be incompatible with it.

My computer guy promised to install it on his computer when he got home. I don't want to have to phone him up every time I need to look up a keyword though. Apparently the release notes for the patch don't give any details about how the keywords work - I can type one in and it either goes to some place seemingly at random, or gives an error.

I tried BATH and it said "Ambiguous", so I tried a few more on a bathing-related theme. It said:

Ambiguous: BATH BATHING BATHROBE
OutsideArea: BATHTOWEL
TooLong: BATHTOWELS
Drawing map for: BATHROOM/BATHROOMS BATHER BATHROBES

I studied the map a bit:

BATHROOM: a place between St. Austell and Truro. There might be a bathroom there, but it seems unlikely.
BATHROOMS: exactly the same place!
BATHROBES: a bit North-West of Truro
BATHER: a field near the other side of the country - between East Grinstead and Cowden.

It certainly seems to be more flexible than the original hacked system, but I'm at a loss of how to figure out what codes to use without the companion program.

Day 2

I tried some more keywords, and was astonished to find that when I put the name of my small town in as a keyword, it actually came to the right place, within half a mile of the town hall. I tried a few more place names of different lengths from a site I found, starting with 'A'.

Ambiguous: ABINGDON ACTON ALFORD ALFRETON ALLONBY ALNWICK ALTON AMBLE ANDOVER APPLEBY ARUNDEL ASHFORD ASPATRIA ATHERTON AYLSHAM AYSGARTH
TooLong: ACCRINGTON ALTRINCHAM
NotFound: ALCESTER ALDEBURGH ALDERSHOT ALNMOUTH ALSTON AMBLESIDE AMERSHAM AMESBURY AMPTHILL ASHBOURNE ASHBURTON ATHERSTON AVONMOUTH AXBRIDGE AXMINSTER AYCLIFFE AYLESBURY
Drawing map for: ASHINGTON
ASHINGTON on my hacked map

Later, my computer guy called me back "I've managed to install the companion program they mentioned. It seems fairly simple - I can click on the map and it tells me some nearby keywords. You can choose from really precisely specified but usually long and practically meaningless keywords - DIDTROSBE for the street I live on, GALMSRAUK or TREVAOUT for the park round the corner, and FRONARFOX for the car dealer I go past soon after I leave the house on the way to your place... or you can insist on a shorter keyword - FROFOX is pretty close to all those, and TEAPMA covers the industrial estate containing the car dealer (and a few others), and finally you can filter to stuff that's in a dictionary... but when one of the dictionaries suggests obscure words like ORESTUNCK, I'd probably recommend the other options for most cases!"

I looked up a couple of places he mentioned. "Oh, you live on that side of town - I thought you'd been speeding the time you got here only 15 minutes after I called you!"

I explained what I'd done trying to look up place names rather than "keywords". I wondered where Ashington actually was...

"Let me check... you can use ALOUDBBW, or SAHAAT - depends which you find more memorable."

I plugged the codes in, and sure enough, it showed me exactly where Ashington is... ALOUDBBW and SAHAAT on my hacked maps "It's really that quick to look them up?"

"Sure, call me back if you need to look up anywhere else."

Day 3

I didn't have anything else I wanted to look up the next day, so I left my computer guy in peace. Instead, I drew a map of all the locations that have already been specified above (will be updated as or when further clues are given). [Moved to Summary section]

Day 4

I need to get my car serviced tomorrow, before I go for a short break this weekend. I was about to call my computer guy to look up the keyword for the car dealer, and another for my normal holiday destination, so I can practice using the hacked maps system to navigate, but then remembered he already gave me a keyword for my car dealer a couple of days ago (the same one he drives past after leaving his house), and besides - I might want to go somewhere different for a break this weekend. I'll call him tomorrow when I've asked my friends for ideas of destinations for my short break. In the meantime, while waiting for my TOAST (Ambiguous) to cook, I looked up where the HONEY I wanted to spread on it had come from.

HONEY on my hacked map

Day 5

The service for the car went well, and the hacked map system seems to work great once you know what keyword to use for each place you need to go to. Although I asked several friends for ideas of where to go this weekend, none of them got back to me (even though a couple of them "liked" the post asking for ideas!). So I figured, what better to find a random location to search than a hacked map system? I went to find TODAYS code, and found a section of map near Ashby-de-la-Zouch. Decision made. I looked up availability on a popular booking website [at time of publication, the availability corresponded precisely to the right part of the screenshot below], and found it was showing 4 hotels. I called my computer guy, asking him for keywords to use for each hotel. He came back with MYBOLFBEN HOWLETHOO GRTPLUBEM CYPHUBACY

MYBOLFBEN HOWLETHOO GRTPLUBEM CYPHUBACY TODAYS

"I'm glad you called back", he said. "I've been digging into the patch folder that I installed, and there's a load of data files. I opened one and it had a list of place names, including all the ones they suggested using to test the system, and another looked like a copy of a dictionary. One of them just had 3 letter keywords with pairs of numbers next to each - keywords like MAN,SOU,OXF,LGW,INV, each with a pair of numbers - probably some kind of co-ordinate.". I typed the short codes into my hacked maps system. Most came to towns beginning with those letters - Manchester, Southampton, Oxford, Inverness. I wasn't sure if I'd see something starting Lgw if I squinted a bit...

"Tell you what", he said, "I'll pop round and have a closer look at your system tomorrow if you like, see if I can figure out how it uses all those files". I reminded him I was going away from the weekend. "I'll call you when I get back."

Day 6

I got to the hotel I booked with no problems, thanks to the keywords my computer guy gave me. It was somewhat before the check-in time, and I wanted to explore the area anyway. I didn't want to bother my computer guy with every idea of a place to visit, so instead I bought a paper map like this from a local shop, and abandoned any attempt to use my hacked maps system other than for returning to the hotel.

I decided to go for a walk during the afternoon. I soon found myself near another group of walkers. One spotted my maps device hanging round my neck, and he was curious why I wasn't using it. As we walked, he seemed fascinated as I described everything that had happened with the hack, and all the things I'd tried with it. As I paused for a short rest, I noticed he'd picked up my map device. "My mate had one of these. The 3-letter keywords are the easiest to use; there's really obvious ones like the place I met my friend when she came to visit - here you go". I took the device back and studied the map.
Location shown when a stranger borrowed my device
It looked kind-of familiar, I definitely recognise something on there. Then I noticed he'd not selected the option to show the labels. I went back to tick it, but the text box was cleared out. "Hey, what did you type in?"

The stranger was nowhere to be seen. Perhaps he wasn't as interested in the conversation as I thought. As the weather started to close in, I finished the walk and returned to the hotel.

Day 7/8

Most of the places I tried to visit either had entry restrictions or looked like they might be too crowded. As I now had a good walking map I went for another walk where I met a beekeeper tending his hives. He told me about how a honeybee will always travel directly between its nest and a source of nectar or honey.

While I was bored during the evening, I entered a few words I'd heard in the conversation into the hacked maps system.

NotFound: HONEYBEES FLOWER FLOWERS BEEHIVE
Ambiguous: NECTAR
Drawing map for: HONEYBEE HONEYNEST

HONEYBEE was hovering around near Fairford, whilst HONEYNEST was way up north not far from the M6 near Warrington.

HONEYBEE and HONEYNEST on hacked maps system

I later discovered there was a HONEYPOT between Worcester and Stratford-upon-Avon. Perhaps the honeybee preferred to be closer to that than its own nest?

I spent the last night at the hotel, and returned home. My computer guy had seemed keen to look at something on my system, but maybe I'll leave it til tomorrow to call him round.

Day 9

I forgot to call my computer guy, but later in the day, he phoned:

"You remember that event you're planning to go to in a few months at the NEC?"

"Sure, what about it?"

"Well I was trying to figure out how the system works, I was looking for keywords around there and, besides the nearby 3-letter code, I found one you might like: LAKSHYSUR - that'll remind you of one of your friends, right?"

I looked it up:

LAKSHYSUR on hacked maps

"Nice, so you figured out how it all works then?"

"Not even close, I'm afraid, just good searching with the companion program. I figured out what some of the files were - there's a Scrabble words dictionary, and an extract from place names used by a crossword solver. Can I pop round this evening to see if I can figure out how your system is using them?"

"I'll order PIZZA (NotFound)... ummm PIZZAS (Ambiguous) ... well, I'll get them even if I can't show you where they might have been on my map!"

Day 10

I woke up with a terrible headache, and tried to remember what happened last night...

I got pizza in, and my computer guy showed up with a crate of beer. It seems to be mostly empty now, and I don't remember seeing him drink much of it. He'd been surprised when he first connected to my maps device, as most of the files he'd found in the patch hadn't been there, but what he found out beyond that... I can't have been paying much attention.

I found a note from him:

So, I found the file the patch author must have used - you were right with that guess of yours that I thought were just drunken ramblings - I got an exact match with the only data file it installs on your device with a file on a government website listing exactly the same set of position references. At least now you know how the 3-letter keywords work when you're just after a general area. Hopefully you'll figure out the rest soon enough. Anyway, I'll let myself out.
PS I've spent enough time on this already - 2 more free calls about this (e.g. if you still need more keywords for specific locations) and after that my normal consultancy fee will have to apply.

I also found a scrap of paper that looks like it might have been important before I used the rest of it for ... other purposes ... during the middle of the night.

Running Monte-Carlo analysis for 26 strings of length 1...
Checked length 1. 0% valid, 100% invalid, 0% ambiguous, 0 outside area

Running Monte-Carlo analysis for 676 strings of length 2...
Checked length 2. 9% valid, 38% invalid, 53% ambiguous, 0 outside area

Running Monte-Carlo analysis for 17576 strings of length 3...
Checked length 3. 15% valid, 85% invalid, 0% ambiguous, 0 outside ar

Running Monte-Carlo analysis for 20000 strings of leng
     ed length 4. 1% valid, 56% inv

Well, I'm not going to use one of my free calls just to call him and ask him to remind me what I was going on about in the middle of the night. I need some COFFEE (NotFound)

Day 10, part 2

As I was gradually starting to sober up, I realised I'd probably upset my computer guy a bit. I must remember to apologise next time I contact him. In the meantime, my computer pinged - incoming message from one of my friends suggesting some 3-letter keywords to try. Astonishingly, most of them actually worked, and they even had links to pretty much the same places my maps system finds... I put all 24 pages into my hacked maps system at once.

Ambiguous: BKJ BIA BSH CLJ HRW LVC LPY LBG STP VIC MRF NWX NXG NRC OXT PMR RDG RMD SHT SGB SBJ UPM VXH WFJ WMB WYB WIJ WIM
NotFound: DNN DKR FLS HXX HOD HRE KCM WIT MLR NQC NHY NWW OXP PNC SHA WFS WFW WYQ
Drawing map for:

It crashed when it tried to draw the map "414 Request-URI Too Long". Still that gives hundreds of other places I can get to without bothering my computer guy (if I even dare call him again).

Day 11

With my the help I got from my friend @Mohirl, I figured out how to make a trip to visit my NAN. It was only about half an hour away; I should do that more often.

Day 12-13

Nothing much to report. When I started to take the rubbish out, I noticed a several pieces of paper with my computer guy's writing on them, and a bunch of computer stuff. They must be notes from the other night. Although there were several of a similar form, one particularly struck me as useful... the final line had been circled several times in red ink:

wik/airUK: 37% valid
wik/UKrs: >99% valid locmatch !! maybe useful - chk src.
-> NR/sc: >99% valid nolocinfo
[drunk genius suggestion "try a government one"]
NaP/AR: >47% valid but many locnonmatch X
NaP/CR: X (nothing relevant)
NaP/FR: >52% valid - locnonmatch X
NaP/MR: X (empty)
NaP/RR: !!!!!!! 100% match !!!!!!!

Day 14-16

I looked in my web browser history, and found the site my computer guy had downloaded the patch from

[for avoidance of doubt, the website is fictional, and the screenshot simulated, but the data is real]

Bring your hacked maps system back to life!

It also had a download page, and acknowledged a bunch of sources including Scrabble, bestforpuzzles, NaPTAN, gridreferencefinder, OpenStreetMap, and Ordnance Survey.

I wish my computer could run the companion program, so I could just click on the map and see all the keywords... and what a shame they focussed on Ely Cathedral for their example. If they'd done Durham, I'd know the right keyword to use for that cathedral already!

Day 17

With the help of my friend @Stiv, I finally understand how the keywords are used, but that still hasn't yet helped me to know how to find the right keyword for a new destination. I decided to call my computer guy and ask him to look up Durham Cathedral for me.

He recognised my number and answered immediately. Before I could get my question out, he said "You've got a nerve calling me up after you drank all my beer last week!"

It turns out the beer was a crate he'd just picked up from the off license for a party this weekend, and he'd only brought it in to avoid leaving it in a hot car. He only noticed it was his beer I'd been drinking rather than my own as he let himself out and found his crate almost empty, and me in no state to answer for myself...

I agreed to get a replacement crate and drop it off at his place tomorrow. Perhaps it'll be better timing to discuss the last couple of maps locations then rather than right now. Maybe someone will get me the right keyword so I don't even have to ask him... I apologised again and hung up.

Day 18 (conclusion)

I got up early in the morning, and headed round the corner to my local off-license. "It's for a friend" I explained to the shopkeeper as I bought the crate of beer.

Just over 15 minutes after entering DIDTROSBE, I pulled up outside my computer guy's house. It's just as well I recognised his car, as I forgot to ask what number he lived at.

He answered the door, saying "Hey, you brought my beer back! Sorry I shouted at you yesterday - you caught me at the end of a bad day. How are you getting on with that maps system?"

"Well, my friend @Stiv has given me some detailed instructions, and I might be able to generate keywords on my own, if I can get Excel to do what it says here..."

"My next job's not for another hour - let me take a quick look - help yourself to a coffee if you want one".

By the time I'd made the coffee, he was furiously typing at his computer.

Half an hour later he handed me a memory stick. "Done better than Excel for you ... just run this, look up the grid reference of where you need to go and type it in, and it'll convert that to a list of keywords you can use."

I thanked him, and after I left, carried on into town to do some shopping before making the drive back home.

I looked up a reference for Durham Cathedral, and entered it into the program to see if my friend's answer could be improved on.

I smiled as I imagined a gas canister blocking the way as you walk around the cathedral. LPGINPATH indeed - that'll do nicely!

I might not have fixed the hack, but at least I've got a usable system now - maybe I'll even be able to install the official companion program when I next upgrade my computer - I wonder if my computer guy will be able to get me a good deal on one...

Summary

The problem is to figure out how valid keywords are processed by my "hacked maps" to generate a UK location. [Originally envisaged as being on my computer, but it's morphed into a "maps device" for narrative purposes!]

3 main characters in the narrative:

  • I have a hacked maps system that only accepts keywords that conform to certain rules. I can enter keywords into it and see where they go, or what error they give.
  • "Computer guy" has a program that can search for keywords near a specified location. He's found "a load of data files", only one of which ("matching something on a government website") is installed on my system, but didn't know how the system was using them, until @Stiv helped me explain it to him.
  • "Patch author" knows exactly how everything works, but cannot be contacted. The release notes were somewhat lacking!

During the narrative so far, the following keywords have generated the error code indicated:

  • NotFound: GERMANY LONDON BRAZIL BATH BATHING BATHROBE ALCESTER ALDEBURGH ALDERSHOT ALNMOUTH ALSTON AMBLESIDE AMERSHAM AMESBURY AMPTHILL ASHBOURNE ASHBURTON ATHERSTON AVONMOUTH AXBRIDGE AXMINSTER AYCLIFFE AYLESBURY HONEYBEES FLOWER FLOWERS BEEHIVE PIZZA COFFEE
  • Ambiguous: SPAIN FRANCE ABINGDON ACTON ALFORD ALFRETON ALLONBY ALNWICK ALTON AMBLE ANDOVER APPLEBY ARUNDEL ASHFORD ASPATRIA ATHERTON AYLSHAM AYSGARTH TOAST NECTAR PIZZAS
  • OutsideArea: BATHTOWEL
  • TooLong: BATHTOWELS ACCRINGTON ALTRINCHAM

The following keywords have valid UK locations which have NOT been specified. Being able to figure out where they are in the UK is the main part of the puzzle. (...) for keywords known to be physically close to each other.

CAYMAN ISLAND DURHAM WARSAW POLAND KENNEL CYPRUS DALLAS JETMOM NORMANTON CLITHEROE BRAINTREE MAXIMUM MINIMUM (DIDTROSBE GALMSRAUK TREVAOUT FRONARFOX FROFOX TEAPMA ORESTUNCK) LGW NAN

Some locations have been specified without stating the keyword - these are each associated with a question below, and form secondary objectives to the puzzle.

The following keywords have valid UK locations which have been stated. Most are shown on the map below (within spoiler tag).

BRASIL ELY YORK OTLEY WHITBY NEWYORK SOUTHEND EDGBASTON SECRET DISMAY KELSYTOUN MARKEH MESUCK STODGY BATHROOM BATHROOMS BATHROBES BATHER ASHINGTON (ALOUDBBW SAHAAT) HONEY (TODAYS MYBOLFBEN HOWLETHOO GRTPLUBEM CYPHUBACY) MAN SOU OXF INV HONEYBEE HONEYNEST HONEYPOT LAKSHYSUR

Summary map:

Map of all clues given at a specified location so far
(Additional clues physically close to TODAYS are on day 5 update.)
(Additional clues physically close to ELY are on screenshots in day 14-16 update.)

Can you figure out how the keywords work?

[Questions in narrative order not necessarily in order of difficulty. An answer which demonstrates a clear understanding of precisely how the keywords work doesn't need to cover every single sub-question - they're intended as hooks to allow partial answers if you figure out part of the puzzle]

  1. Tell me the "most obvious" thing I could type into the box to actually locate a few towns. [Answered by @Mohirl]

  2. Tell me where some (or all!) of the places from the previous episodes end up now (i.e. CAYMAN ISLAND DURHAM WARSAW POLAND KENNEL CYPRUS BRASIL DALLAS) [Answered by @Stiv]

  3. Where did the person who provided the patch claim their mom went on a jet-ski (JETMOM)? [Answered by @Stiv]

  4. Where would I find NORMANTON, BRAINTREE and CLITHEROE? [Answered by @Stiv]

  5. About how far is MAXIMUM from MINIMUM? [Answered by @Stiv]

  6. If this new patch is all it's cracked up to be, I don't want to merely bookmark the correct location of Durham, but specifically Durham Cathedral How close can you get me to it? [answered within 130m by @Stiv]

  7. Where do I live (for the purposes of this question!) - see Day 2 and Day 11

  8. Where does my computer guy live? - see Day 2 [Answered by @Stiv]

  9. What brand of car do I drive? - see Day 4 and Day 2 [Answered by @Stiv]

  10. What did my hacked maps show me when I entered LGW? [was trivial to look up from @Mohirl's answer, but @Stiv was first to directly state it]

  11. What did the stranger enter into the system to display the location shown on Day 6 [Intended answer not yet given. It was (unexpectedly at time of asking, but anticipated shortly afterwards) real-life ambiguous... @Stiv has provided one answer that arguably should be correct, and another that is correct but unintended - I'll address this in a retrospective after awarding the green tick later today]

  12. Can you give more specific details about why some keywords return Ambiguous rather than NotFound? [Answered by @Stiv]

Summary of correct answers so far:

[@Mohirl's partial answer comprehensively covered the approximate location of almost all 3-letter keywords, and gave some initial ideas about how those might be combined to make 6-letter and 9-letter keywords, but after a few more hints and updates, @Stiv has pulled out a tremendous effort to provide almost all the remaining missing details - the only thing outstanding is "where do I live"]

Hints:

I can give interactive clues - let me know in comments, or a chat associated with this question what places I should ask "the computer guy" to look up tomorrow, or what keywords to type in to my hacked maps and see where they end up. (NB I'm reluctant to call him after the events of days 9/10 - give me a good reason for any places I need to look up)

A lot of the "flavour text" is indeed just for flavour, but there are definitely clues there too.

As a first step towards understanding the answer, consider what some of the 3-letter keywords might represent.

There are several different reasons for a "Ambiguous" result, but all are precisely defined. The "patch author" had to specifically check for each one. Some of the valid keywords may be ambiguous in real world usage, but all have a very precise meaning in terms of the system. If someone starts on the right track, I'll try to signal any missing details.

When the basics are worked out, consider why this puzzle dropped the word "Google" from the title, and none of the example map screenshots use Google - perhaps Google Maps doesn't support the co-ordinate system needed for a full answer?

Pay close attention to the extracts of the release notes from the patch author.

Look at the "scrap of paper" from day 10. Given a list of 3-letter keywords, how might you make a list of 2-letter keywords most of which are "Ambiguous"? What do the non-ambiguous ones represent?

Screenshot in day 14-16 update has plenty more keywords visible when expanded to full size.

A triangle isn't the only shape uniquely defined by 3 points.

Taking the train is often a faster way to get directly between train stations than driving, even if the train does stop at other places along the way.

$\endgroup$
  • 2
    $\begingroup$ I feel like I should try this not because of bounty but also because it is beautifully crafted.+1 $\endgroup$ – Lakshay Sura Jul 29 at 17:10
  • 3
    $\begingroup$ Dear friends on PSE, I'm looking for ideas for destinations for a short break in the UK this weekend. I (or more precisely, the character I'm playing in this puzzle) will select one of them as a holiday destination for the weekend. This isn't a trick question; right now, I'm looking for actual places (just give a real address, name of resort, link to an online map, or similar), not keywords to try, at least one of the ideas given will be incorporated into the "day 5" update tomorrow. Please comment (or add a chat to contain several people's suggestions if you know how). $\endgroup$ – Steve Jul 30 at 13:09
  • 3
    $\begingroup$ I note that the cumulative updates have made this quite long now... in next update later today, I'll reorganise things to add a summary, and to move the suggested partial questions to the end. $\endgroup$ – Steve Aug 1 at 9:42
  • $\begingroup$ I guess Otley seems to break the rule. It doesn't have any prominent cathedral. Do correct me if I'm wrong. $\endgroup$ – Lakshay Sura Aug 1 at 10:08
  • 2
    $\begingroup$ @Steve I would love to post an answer for this question, as you've clearly gone to some lengths in setting it up. However, despite knowing what most of the puzzle is all about and how to proceed with it, I'm missing something fairly standard - I cannot for the life of me replicate the original ASHINGTON result (no matter what coord system I use I end up near Nottingham rather than out in the sea). Without being able to replicate that most basic part of the puzzle I fear I will not be able to formalise the solution at all! $\endgroup$ – Stiv Aug 12 at 20:54
11
+200
$\begingroup$

A lot of thought and prep time has clearly gone into this puzzle and all it subsequent updates. I will attempt to do its complexity justice by answering your 12 summary questions in order, providing a few diagrams along the way...

  1. Tell me the "most obvious" thing I could type into the box to actually locate a few towns.

As discovered by @Mohirl, the mappings in this puzzle all rely on the 3-letter codes for UK National Rail stations.

In the simplest case, entering the 3-letter code for a specific station will enable you to pinpoint that location exactly. However, longer keywords enable you to identify other locations based on a few specific rules...

- If a 6-letter keyword can be broken down into two valid 3-letter codes, the given location will be geographically halfway between the two. (See STODGY for Edgbaston as a good example...)

- If a 9-letter keyword can be broken down into three valid 3-letter codes, the given location will be found at the centre of a circle which passes through all three of the points.

- If a 3-letter code can be represented uniquely by just its first two letters (namely, that if the station's code is ABC, there is no other station whose code is of the form ABx) then these two letters are sufficient in the keyword without needing to use all three letters. This enables the use of keywords with a number of letters other than 6 or 9.

My roundabout process for solving these was to find latitude-longitude coordinates for the railway stations using Google's 'My Maps' (easier to pinpoint them here), enter these into gridreferencefinder.com to translate these into Northings and Eastings, and use these values in any calculations in Excel.

  1. Tell me where some (or all!) of the places from the previous episodes end up now (i.e. CAYMAN ISLAND DURHAM WARSAW POLAND KENNEL CYPRUS BRASIL DALLAS)

CAYMAN (Carntyne CAR -- Manchester Piccadilly MAN) = just off the A591 near Bassenthwaite, Cumbria.

ISLAND (Isleworth ISL -- Anderston AND) = a quarry south of Bacup in Lancashire.

DURHAM (Durrington-on-Sea DUR -- Hamworthy HAM) = the Solent, between the Isle of Wight and Lee-on-the-Solent.

WARSAW (Ware (Herts) WAR -- Sawbridgeworth SAW) = just east of Hunsdon, East Herts.

POLAND (Polsloe Bridge POL -- Anderston AND) = a spot in the bay by Conwy, Wales.

KENNEL (Kendal KEN -- Nelson NEL) = the moor south-east of High Bentham, North Yorkshire.

CYPRUS (Crystal Palace CYP -- Ruswarp RUS) = a field east of Scredington, Lincolnshire.

BRASIL (Brora BRA -- Sileby SIL) = near Upper Hindhope, Scottish Borders.

DALLAS (Dalmally DAL -- Llansamlet LAS) = the Irish Sea, south-east of the Isle of Man.

  1. Where did the person who provided the patch claim their mom went on a jet-ski (JETMOM)?

The keyword JETMOM is a 3-point reference (it's not as simple as JET+MOM): JEQ (Jewellery Quarter), TMC (Templecombe) and OMS (Ormskirk). Plotting the Northings and Eastings of these points on a circle, we find that the centre-point lies about 7 km out from the Strumble Head lighthouse.

enter image description here

  1. Where would I find NORMANTON, BRAINTREE and CLITHEROE?

NORMANTON (Normanton NOR -- Manchester Piccadilly MAN -- Tonbridge TON) = near Hanging Houghton, Northamptonshire.

BRAINTREE (Brora BRA -- Ingatestone INT -- Reedham (Norfolk) REE) = the Irish Sea, between Ramsey (Isle of Man) and Gosforth.

CLITHEROE (Clifton (Manchester) CLI -- Theale THE -- Rotherhithe ROE) = west of Barleythorpe, Rutland.

  1. About how far is MAXIMUM from MINIMUM?

MAXIMUM (Maxwell Park MAX -- Imperial Wharf IMW -- Umberleigh UMB) = White Man's Dam, North West England. Northing 393997.1472, Easting 345048.4156.

MINIMUM (Milliken Park MIN -- Imperial Wharf IMW -- Umberleigh UMB) = just east of Knowsley, North West England. Northing 396287.4108, Easting 344591.9804.

Conveniently, these two locations are very close to each other! About 2.30km apart:
enter image description here

  1. If this new patch is all it's cracked up to be, I don't want to merely bookmark the correct location of Durham, but specifically Durham Cathedral How close can you get me to it?

At this stage it's worth noting the significance of the piece of paper found on Day 12-13, and specifically the importance of the last line, "NaP/RR: !!!!!!! 100% match !!!!!!!". From the context of this piece of paper it seems that the 'computer guy' has been attempting (and struggling) to identify the precise list of stations used to generate the keywords - each line represents a new data source and how closely it corresponds to the source of the mapping tool's keywords. This last line circled in red (and exclamation marked to oblivion!) represents NaPTAN/RailReferences.csv (NaP = NaPTAN; RR = RailReferences), a file freely available to download from the data.gov.uk website, which contains not only a full list of station codes, but also - get this - their Eastings and Northings...!! (If I had spotted this sooner I could have saved myself a lot of hassle with alternating between Google Maps and gridreferencefinder.com! But never mind...)

Using this spreadsheet, it suddenly becomes much easier to find a suitable keyword for a given location. Simply ('simply'...!):

1. Find the Northing and Easting of the target location.
2. Enter these in every row of the spreadsheet in two columns at the end of the file.
3. Calculate the difference between the Eastings of each station and the target, and the difference between the Northings of each station and the target.
4. Use Pythagoras' theorem (a^2 + b^2 = c^2), realising that the difference in Eastings and the difference in Northings comprise the two shorter sides of a right-angled triangle - the hypotenuse (which we are calculating here) is the distance between the station and the target.
5. Sort the file by this calculated hypotenuse length.
6. Add a column which compares this length with that of the station 2 further along in the ordered list. We are trying to find three stations that have as similar as possible a distance to the target, in order to draw the most precise circle.
7. Examine one by one the station trios picked out as having the shortest discrepancy in distance between their three calculated hypotenuses. Discard any where the same station code appears more than once - there are duplicates in this list (e.g. Clapham Junction CLJ appears 3 times in the dataset, possibly due to having multiple entrances and both a National Rail station and a London Underground station).
8. Calculate the keyword as before, and bingo - you have your answer!

In the case of Durham Cathedral, if we set the cathedral's coordinates as being those which show up by default in a Google Maps search (Easting 427349, Northing 542117), the three non-duplicating stations which produce the smallest discrepancy in distance to the cathedral are:

Chorleywood CLW -- Derby Road (Ipswich) DBR -- High Wycombe HWY

which yields the keyword CLWDBRHWY (ordering them alphabetically) and can be seen plotted on the following map:

enter image description here

This circle is centred on a location roughly 114m from the heart of Durham Cathedral, as pinpointed by Google Maps, and an even shorter distance to the nearest boundary wall!

enter image description here

  1. Where do I live (for the purposes of this question!)

A little more work to be done on this question still...
Looking up one of the keywords mentioned as being in the same town as the OP's house (specifically a car dealership), FRONARFOX is equivalent to Frome FRO -- Narberth NAR -- Foxfield FOX, which works out as a location in the Harlescott area, just north of Shrewsbury:

enter image description here

The dealership in question is 'Shrewsbury Audi':

enter image description here

  1. Where does my computer guy live?

The computer guy mentions that DIDTROSBE is the keyword for the street on which he lives. This is equivalent to Didcot Parkway DID -- Trowbridge TRO -- Starbeck SBE, which works out as Waincott in the Harlescott area, just north of Shrewsbury.
enter image description here

  1. What brand of car do I drive?

Since the narrator gets their car serviced at the location identified for Q7, they must drive an Audi!

  1. What did my hacked maps show me when I entered LGW?

LGW is the railway code for Langwathby, near Carlisle - not (as you may expect) London Gatwick Airport...

  1. What did the stranger enter into the system to display the location shown on Day 6?

...and speaking of airports, the location on Day 6 is Heathrow Airport. The 3-letter code typed in by the stranger would have been HTR, the code for Heathrow Airport Central Bus Stn (Rail-Air).

NB Not HXX, the code for Heathrow Airport Terminals 1, 2 and 3, since although the HXX code does appear in many station code sources it does not appear in the dataset specifically used by the OP (as clued in the account of 'Day 10, part 2').

  1. Can you give more specific details about why some keywords return Ambiguous rather than NotFound?

Let's explain using an example.

SPAIN is an ambiguous keyword because although SPA is Spalding, there is more than one station with a code of the form 'INx' (e.g. Invergowrie ING, Inverness INV, Inverkip INP...). Alternatively, although AIN is Aintree, there is more than one station with a code of the form SPx (e.g. Shepherd's Bush SPB, Shepherd's Well SPH, Shepley SPY...).

Since there is more than one way to form the keyword it is not possible to derive a unique location (in contrast to those which are NotFound, which match to no valid locations at all).

| improve this answer | |
$\endgroup$
  • $\begingroup$ Brilliant! I definitely wouldn't have got the rot13(pvepyr) bit, that makes a lot more sense $\endgroup$ – Mohirl Aug 13 at 10:59
  • $\begingroup$ An impressive and comprehensive answer; almost perfect. However, you missed a couple of details relating to Q7 - "when I put the name of my small town in as a keyword, it actually came to the right place". When later talking about "town" to a resident of the general area, however, that means the big town you identified. Also, in Q11, the ..X code is correct by most reasonable definitions but was not the intended answer - a source of some frustration when I re-checked after day 6 update. See the shorthand in day 12-13 update for as close to a direct link as I dared, and also day 10 part 2. $\endgroup$ – Steve Aug 13 at 11:34
  • $\begingroup$ Another way to describe rot13(gur prager bs n pvepyr juvpu cnffrf guebhtu nyy guerr bs gur cbvagf) is rot13(ng n cbvag na rdhny qvfgnapr sebz nyy guerr bs gur cbvagf). This may help you to rot13(svaq fbzrguvat nccebcevngr gb qb va Rkpry gb vqragvsl guerr fgngvbaf nyzbfg gur fnzr qvfgnapr sebz n fcrpvsvrq cbvag.). -- PS These comments are from memory - I'll be at the computer with all the answers on it in a few hours, so will re-check everything then. $\endgroup$ – Steve Aug 13 at 11:46
  • $\begingroup$ @Steve I realised that about Q11 after posting! Will correct it tonight and pick up on a few of your other points too... $\endgroup$ – Stiv Aug 13 at 11:55
  • 1
    $\begingroup$ @Steve I've just pinpointed your data source thanks to your comment above. THAT would have saved me a LOT of faff with coordinate translations! Sigh... Ah well, I'll have another crack at it tonight when I have spare time... $\endgroup$ – Stiv Aug 13 at 13:25
10
+100
$\begingroup$

Partial answer

Posting this as a partial because I don't have time to devote half enough time to the wonderful array of clues provided, but I think the secret is

UK railway station three letter codes.

Take for example:
EDGBASTON: Comes up to a place near the M5, south of Eastington. If you actually want Edgbaston, keyword STODGY will get you there, specifically to the golf course.

keyword STODGY will get you there

Stodgy breaks down into STO (South Tottenham) and DGY (Deganwy). About midway between the two is Edgbaston

Meanwhile

EDGBASTON: EDG is Edge Hill, BAS is Bere Alston, and TON is Tonbridge. The centre of the triangle the three form looks like it might just be "a place near the M5, south of Eastington"

This explains the words rejected as "too long"

They're all more than 9 letters, which is the longest word you can make from three 3-letter code.

The reason you can get more accurate locations from longer words is that

You're triangulating from three locations instead of two.

The three letter keywords are:

Just individual stations, many of which are in towns/cities beginning with those three letters

I'll try to add more later, but hopefully someone can build on this if it's right.

| improve this answer | |
$\endgroup$
  • $\begingroup$ You're definitely on the right track... however, keywords of several other lengths are valid, and rot13(gur pbqrf nera'g nf havirefny nf V nffhzrq jura frggvat gur chmmyr. Gur yvfg V'z hfvat qvssref sebz jvxvcrqvn, fb abj V unir na nqqvgvbany harkcrpgrq fgrc bs gelvat gb tvir pyhrf gb JUVPU yvfg bs pbqrf ner orvat gerngrq nf pnabavpny sbe gur checbfrf bs guvf chmmyr) $\endgroup$ – Steve Aug 5 at 13:34
  • $\begingroup$ Pun intended in the first sentence? Sorry for making it more complicated - I went with Wikipedia as that's the first option I found, and it also included map links. But if there's a more authoritative source that also allows for the other lengths, I'd say it's fine as is to treat that list as canonical $\endgroup$ – Mohirl Aug 5 at 14:51
  • 2
    $\begingroup$ Pun very much intended (as well as the first time it appeared as a hint). It's not you making it more complicated, but the data. The first time I noticed any inconsistencies was rot13(Whfg nsgre V'q qbar gur qnl fvk hcqngr.). Oh, and obviously, as the only person to make any significant visible progress towards a full answer, you'll get the [first?] bounty, $\endgroup$ – Steve Aug 5 at 15:10
  • $\begingroup$ A further update notes the differences between rot13(gur pbqrf ba gur jvxvcrqvn cntr naq gur barf erpbtavfrq ol gur flfgrz. Gur qvssreraprf ner sne srjre guna V'q srnerq jura V svefg fcbggrq n qvssrerapr. Pbqrf erpbtavfrq ol gur flfgrz ohg ABG ba gur jvxvcrqvn cntrf ner abg fubja.) $\endgroup$ – Steve Aug 5 at 15:52
  • $\begingroup$ Just occurred to me that there may be a minor misunderstanding with your comment "The reason you can get more accurate locations from longer words is that..." - all keywords individually have about equal accuracy, but as "computer guy" has a search mode that will only show him shorter keywords, he considers them to "cover" the area where the precise point identified by the short keyword is nearer than any other similarly short keyword. Will insert something along these lines as a clue into next update. $\endgroup$ – Steve Aug 10 at 15:15
4
$\begingroup$

Retrospective

@Stiv has successfully figured out almost every detail of the system. I was beginning to worry nobody would get there. I'll not repeat what has been laid out in such detail in that excellent answer.

Having spent more than a month on this (about 2 weeks prep time, and more than 2 weeks since the puzzle went live), I hope the community will allow me to indulge in a short(ish!) retrospective (in the form of an "Answer" in order to allow the question to remain "in character").

This assumes you already know all the answers, so assume major spoilers on any link and spoiler block!

Source code

First, for anyone that wants to inspect the inner workings in more detail, or simply play with this or a similar system some more the C# source code is now public.

This replicates the functionality (except the UI) in the narrative of the "hacked maps" system, the "companion program" and the equivalent system that my computer guy made for me when I was able to give him @Stiv's detailed explanation of how the codes work.

I was impressed that @Stiv came up with almost exactly the same algorithm to run in Excel as the program uses internally, albeit without the secondary condition that reduces the number of locations that need to be constructed.

Additional ("hidden") clues towards how the scheme worked.

  • a honeybee will always travel directly between its nest and a source of nectar or honey

HONEY, HONEYBEE, HONEYPOT, HONEYNEST are all on a perfectly straight line on the map.

This hints at the fact that

There are two relevant points that are equidistant to every single one of those, specifically HON and EYN

  • FRONARFOX is near FROFOX, which hints that

adding a third point doesn't necessarily shift the position by a long way, ruling out simple averaging schemes etc... especially once you figure out where FRO and FOX are.

  • OutsideArea

Perhaps I should have thrown in a couple more... the number of otherwise perfectly great clue-words that got thrown out due to this error, I had plenty to spare. I'd expected this to come up a bit more if I got other people's suggestions of codes to try. The mere existence of this as a separate error was supposed to hint at the "equidistant from all 3 points" or "centre of circle" way of interpreting the codes. Obviously, nobody can figure out the exact criteria for this from only a single clue - anything that doesn't correspond to the usual set of a 100km x 100km grid squares was "OutsideArea" (see lines 250-273 of the source code for exact details).

  • BATHROOMS and HONEYNEST

both revealed a "hidden third letter" of a two-letter code when compared to similar keywords that were presented... and the first in particular reveals 'OM' as a valid 2-letter code, which formed part of the 'JETMOM' target, hinting at how that was constructed.

  • specific co-ordinate scheme

was hinted at so heavily, I'm sure most people picked it up... the removal of "Google" from puzzle title, the "Summary map" being on the OS view from gridreferencefinder.com, and even going out to buy a physical OS paper map on day 6. Never mind the mentions I gave of it in my contributions to the previous 2 puzzles... of course some may have suspected it was a red herring, which brings me to

  • airports!

soon after discovering LGW was a valid station code for somewhere totally different to Gatwick Airport, even before the puzzle was ready, I plotted to set up a diversionary path. It was less successful than I'd imagined, probably in part because it contradicted too many other clues by then. I half-heartedly tried to use the narrative to set up the thought that there may be multiple data files so that people might think it was "airport codes AND[some other code set]" and at least post a partial suggesting airport codes formed part of the answer.

I'd cross-checked

lists of IATA airport codes against my program, and by implication the NaPTAN data set I was using, finding a large number of code matches, of which about a dozen corresponded to the same cities, and 'LHR' was the jewel in the crown, actually corresponding to a location right in the middle of the airport itself (in the data set I was using) so someone would give a right answer for the wrong reason... and after a moment of head scratching investigate further to find a different set of 3-letter codes that have SOU as Southampton, MAN as Manchester, LHR within Heathrow Airport, etc. Just after posting day 6 update I finally thought to double-check this intended solving path would work, quickly double-checking the wikipedia pages for exactly which station has the 'LHR' code and that the correct answer was accessible via that route... and the entire plan for one intended solving path fell apart (yes, HTR is a correct answer too, within 80m of LHR in that data file, but it comes only later after the whole station codes thing has been worked out), which brings me to:

Lesson learned - double check multiple data sources

I'd assumed that

any reasonably official source would be basing their list of valid 3-letter codes for stations on the same ultimate primary source, and therefore picking one up that included those codes with precise locations was "better" than a list that either didn't include locations, or needed them looking up in a different database (e.g. postcodes).

A week or two before the puzzle went live, I'd checked

several of the codes being used against wikipedia and other sources, and when they all matched, I assumed without checking that every remaining code was a match, and concentrated more on finding good keywords to use in the initial phases of the problem.

The problem could have been fixed

I'm still kicking myself for failing to remember until it was too late that

LTN (Luton Airport Parkway for trains, Luton Airport for planes) was a good alternative to LHR (which for reasons I can't fathom is shown instead of 'HXX' in the data file I was using). I'd rejected it initially because the difference in marker position between the station and the airport would be more apparent unless it's zoomed out too far to see the airport (whereas with Heathrow, I just zoomed out far enough that the words "Heathrow Airport" weren't visible), and also because "everyone" knows Heathrow Airport, but I could have switched this screenshot in as the location it showed after the stranger typed something, and then done the check I eventually did on day 10 to see how extensive the damage was.
LTN from the hacked maps system or LTN from the hacked maps system

Instead, because I didn't know

how extensive the differences were, having failed to cross-check those data sources earlier, I ended up starting down the somewhat unsatisfactory path of trying to force answerers to find the exact same data file I'd used to construct the puzzle; something that wasn't planned for originally, so I'd not thought of good hints for it.

A TFL document I found around that time Three Letter Abbreviations caused me particular concern

when trying to find out about differences between the relevant sets of 3-letter codes. How many other sets of almost, but not quite compatible codes could there be, and exactly which one did I have?

It was only a few days later that I discovered

the codes shown in Wikipedia do in fact match >99% so there'd been no need to panic.

The "correct" data file

was only supposed to be necessary for the last few metres of precision, and indeed, @Stiv was successfully able to decode everything else using different locations sources.

Amongst other things, this was ensured by

making sure every 3-part keyword had a significant distance between the three points used to triangulate the final position. In effect, this means that, as viewed from the target point, each of the source locations is at an angle rather more than 5° (or 0.1 radians) from each other, and so each 1m of position difference on a source point will at most move the constructed destination by 10m, but more often the source points are well spread out, and a 1m position difference in the source will cause up to around 1m in the target. I'd originally had an even stricter limit, but loosened it to allow codes such as ASHINGTON to work (it originally gave the "badly formed" warning, but didn't look badly formed enough to me to reject it, especially being so far out at sea that even the last few kilometres of precision didn't matter).

Interactivity?

Despite practically begging for someone (anyone?) to suggest either keywords or UK places to look up, absolutely nobody did. I'll try to remember not to ask for "audience participation" again in any future puzzles... it was intended to demonstrate that

the system really could be used to look up keywords for absolutely anywhere within the valid area...

The search for good clues

There were two basic approaches. Initially, I simply typed in lots of possible keywords to see which were valid, and where they ended up.

YORK was found fairly quickly in this manual search, after

I'd implemented the code-shortening mechanism. Initially this gave "ambiguous" for any 6-letter keyword that could be expressed either as 3+3 or as 2+2+2, but that ruled out way too many previously-valid keywords, and in particular most of the "clues used in previous puzzles in the series", and also felt like it would be practically impossible to solve.

It soon became apparent that

almost all the valid keywords being found were of 9 or 6 letters, with most of the rest either 8 or 5 letters. Fairly early in the preparatory stages, I implemented a monte-carlo analysis to figure out how many keywords I should expect to be valid.

More output of that (quoted to a slightly higher precision than the scrap of paper in the question!):

Checked length 0. 0.0% valid, 0.0% invalid, 100.0% ambiguous, 0.0% outside area
Checked length 1. 0.0% valid, 100.0% invalid, 0.0% ambiguous, 0.0% outside area
Checked length 2. 9.3% valid, 37.7% invalid, 53.0% ambiguous, 0.0% outside area
Checked length 3. 14.6% valid, 85.2% invalid, 0.2% ambiguous, 0.0% outside area
Checked length 4. 0.7% valid, 56.1% invalid, 43.2% ambiguous, 0.0% outside area
Checked length 5. 2.5% valid, 81.2% invalid, 16.3% ambiguous, 0.0% outside area
Checked length 6. 2.8% valid, 70.5% invalid, 26.7% ambiguous, 0.0% outside area
Checked length 7. 0.2% valid, 83.0% invalid, 16.7% ambiguous, 0.1% outside area
Checked length 8. 0.5% valid, 95.2% invalid, 4.2% ambiguous, 0.2% outside area
Checked length 9. 0.3% valid, 99.6% invalid, 0.0% ambiguous, 0.1% outside area

Although there was a low percentage

for 9 letter keywords, this was compensated by the sheer number of 9 letter keywords available. 7 letter keywords were the worst, as almost everything I tried either came back "invalid" or ambiguous.

During the preparation time, I implemented an automatic search, and input into this

a copy of /usr/dict/words I downloaded, and a various lists of place names from bestforpuzzles.com. This gave copious amounts of 6-letter keywords, several usable clue words of different lengths, but 4 letter and 7 letter words were incredibly rare.

I also wanted to use a few

a 6 letter keywords split as 2+2+2. These were even rarer than the 7 letter ones. I generated a complete list of every possible combination of 2+2+2, and ran a dictionary search on it - nothing. Repeating a similar search now, using all of the dictionaries I've since downloaded, I get this output:
WARNING: Badly formed
FYLFOT : SP0907972936(409079.0,272936.8): FYSLFDOTF Circle: Ferryside, Lingfield, Otford
OutsideArea: IGNIFY OMNIFY OMNIUM QUOTUM URNING
Drawing map for: FYLFOT
https://gridreferencefinder.com/?gr=SP0907972936|FYLFOT|1&v=r&labels=1
https://gridreferencefinder.com/osm/?gr=SP0907972936|FYLFOT|1&v=r&labels=1

I didn't consider those usable, which is why I eventually just stuck with the (manually constructed) keyword JETMOM for that purpose.

Similarly, for 7-letter keywords, this is what I was originally working with:

[Redacted NotFound/Ambiguous, it was many hundreds of words long]
OutsideArea: BERNINI CRANIUM DIGNIFY IMBRIUM MAGNIFY MOONING OMNIBUS STONING VAUNTED YOUNGER YOUNGLY
Drawing map for: JERKING MAXIMUM MINIMUM OBEYING QUEUING QUOTING UNSOUND VANILLA https://gridreferencefinder.com/?gr=NY4421304689|JERKING|1,SJ4505094003|MAXIMUM|1,SJ4459496293|MINIMUM|1,SH6031976220|OBEYING|1,SH6017795077|QUEUING|1,SH9445587395|QUOTING|1,SP4650403019|UNSOUND|1,TA3696795369|VANILLA|1&v=r&labels=1
https://gridreferencefinder.com/osm/?gr=NY4421304689|JERKING|1,SJ4505094003|MAXIMUM|1,SJ4459496293|MINIMUM|1,SH6031976220|OBEYING|1,SH6017795077|QUEUING|1,SH9445587395|QUOTING|1,SP4650403019|UNSOUND|1,TA3696795369|VANILLA|1&v=r&labels=1
Code or co-ordinates:englandplaces.txt,7
[Redacted NotFound/Ambiguous, each a few dozen long - this was the list I'd compiled from the 'bestforpuzzles' site]
Drawing map for:
https://gridreferencefinder.com/?gr=&v=r&labels=1
https://gridreferencefinder.com/osm/?gr=&v=r&labels=1
In other words, not very many, and most of those were not very usable. I think I'd intended to introduce one or more of UNSOUND, JERKING, OBEYING later in the puzzle, but kind-of forgot.

Searching world cities finally came up

with URUMCHI,OCOTLAN,NEWYORK, and I kicked myself very hard for not spotting NEWYORK before I'd done such an extensive search.

For some reason, even though I'd downloaded it at the same time as

the NaPTAN data file, it didn't occur to me to process the NTPG data set for possible place names. YORK is still the only possible 4 letter place name that translates as a valid code, but NEWYORK (also a uk place name!) is joined by LLOYNEY and JESMOND. https://gridreferencefinder.com/osm/?gr=NY7976509880|YORK|1,SP2749807371|LLOYNEY|1,SC9832043980|NEWYORK|1,SP7780040588|JESMOND|1&v=r&labels=1

I'm sure it became obvious as I added more (perhaps too many!) clues, but I wanted the initial puzzle post to avoid disclosing the fact that

4 and 7 letter place names are so rare.

A lot of the rest was just trying various word lists, looking for patterns, and expanding on them where possible. For one particular pattern, HONEYBEE, HONEYPOT, HONEYDEW, HONEY was the initial set I was planning to use before I constructed the dual-purpose HONEYNEST.

The actual output when I type in co-ordinates or a grid reference looks something like this:

Code or co-ordinates:NZ273421
Including 3-codes...
INPLPGATH(10.7599239156583) : NZ2730942094(427309.1,542094.2): INPLPGATH Circle: Inverkip, Llanfairpwll, Atherstone
Alternative forms:INPLPGATH/INPATHLPG/LPGINPATH/LPGATHINP/ATHINPLPG/ATHLPGINP
HVNPRSLLJ(25.1577464405969) : NZ2727442098(427274.9,542098.8): HVNPRSLLJ Circle: Havenhouse, Prees, Llandudno Junction
Alternative forms:HVNPRSLLJ/HVNLLJPRS/PRSHVNLLJ/PRSLLJHVN/LLJHVNPRS/LLJPRSHVN
RKTELOKMK(30.3520371149044) : NZ2727042092(427270.6,542092.5): RKTELOKMK Circle: Ruskington, Elton & Orston, Kilmarnock
Alternative forms:RKTELOKMK/RKTKMKELO/RKELOKMK/RKKMKELO/ELORKTKMK/ELORKKMK/ELOKMKRKT/KMKRKTELO/KMKELORKT/KMKELORK
CCCTYLKET(32.4548051673883) : NZ2733142107(427331.7,542107.1): CCCTYLKET Circle: Criccieth, Tyndrum Lower, Kettering
Alternative forms:CCCTYLKET/CCCKETTYL/TYLCCCKET/TYLKETCCC/KETCCCTYL/KETTYLCCC
...
The "Alternative forms" list also includes shortened forms, and is pre-checked to ensure that none of the suggestions would come back as "Ambiguous"... so most of the clues given for specific places were from scanning down the list until I found something meaningful or at least pronounceable and "not too ugly". It was a shame I never got to do that with a user-suggested location. A "dictionary search near specified location" function was less useful than I'd thought it would be (nearest 3 Collins Scrabble Words to "computer guy's" house being ORESTUNCK, RILLED and TRELLISES, with the closest of those about 3km away).

In the end there were a lot of unused or discarded clues, and especially potential clues of length 6 (unless I was targetting a specific position I was spoiled for choice!)

Real-life update

A new valid code has been created since this puzzle was launched (although Wikipedia hasn't caught up yet). If I'd noticed before today I might have done an in-character "system update" and then BACKTRACK OutsideArea...

Where do I live (for the purposes of this puzzle)

This is a place holder section to be used to elaborate on any "spare" clues not identified by whoever posts a correct answer to this part. It's mostly independent of the rest of the puzzle.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.