Mikalye’s Magic Guide to Puzzles – An entirely personal view

Change Log - Whats new in version 2:

1) I added many of the tools suggested in the comments to the references section
2) I added a new section on metapuzzles
3) I added a new section on backsolving.

Change Log - Whats new in version 3:

I have updated the references section.

Change Log - Whats new in version 4:

I have updated the references section.
I have added the new short section on "The goal of the puzzle-writer"
I have added a section on puzzlehunts

Introduction

Look, Zelghadis’ magic guide to puzzles is the original and clearly the place to start. If you haven’t read that then stop reading this right now and go read that first. You can find it here: (http://www.steamgifts.com/discussion/TLR8m/zelgh-magic-guide-to-basic-sg-puzzling-update-june-2015). However, there have been enough questions, particularly around web-based puzzle solving resources, that I felt that there was room for another, perhaps more idiosyncratic view of puzzle writing and solving. As part of this, I really wanted to lay out how I construct my puzzles and my personal views as to what makes a puzzle elegant versus inelegant.

As part of that, while I am writing this, I will try to create some puzzles to illustrate the various points that I am making, and occasionally reference some external puzzles. So there will be several puzzles to solve within this document. Do note however, that as I am creating these on the fly, most of these puzzles may be poor (again, they were written as I type, with no real time wasted on actual thinking). This is meant more as a way of thinking about or driving a puzzle creation than anything else.

Arguably, the most useful part of this for most people will be the tools and references that I link to in the below text.

On Puzzles

A puzzle, in essence, is a problem. A problem starts with given conditions and asks you to derive a particular result. Finding the path from what’s given to what you are being asked for is problem solving. Why do I solve puzzles? Because I am addicted to that AHA moment. There is a real feeling of joy at the moment that the problem solution or some other significant realisation becomes clear. That is my drug.

A Quiz is not a Puzzle

Many of the “puzzles” that appear on SG are not puzzles at all. They are quizzes. They test your ability to know something, or to look something up. They require mad search engine skills. There is nothing wrong with quizzes. I like quizzes. I have written some quizzes. My History of Computer Role Playing Games (http://www.steamgifts.com/discussion/8ayLm/mikalyes-0th-puzzle-of-2016-a-history-of-computer-role-playing-games) was such a quiz. It was more about reading comprehension, and about searching for facts about computer RPGs, than any form of puzzle. The bulk of jeffhowe’s wonderful March 2016 SCIENCE Event consists of quizzes, including my own contribution. Not all of course. Zelghadis does not do quizzes, and so provided a puzzle rather than a quiz (http://www.steamgifts.com/giveaway/W9izh/postal-2)

Why is a quiz different than a puzzle? Because it lacks the AHA moment that is the puzzler’s fix.

The golden rule

A puzzle can hide stuff from you, it can be infuriating, it can lead you down blind alleys, but it should play fair. If a puzzle has multiple steps, then each step should be clued, or motivated. So if you get given a pile of binary text and then convert that to ASCII characters, and then you are expected to recognize that the answer if read backwards in Lithuanian gives the name of a 16th century Polish botanist who once created a cipher, then the puzzle is broken.

If the puzzlemaker has given clues to what he/she needs you to do, then all is fine. If the next step is “Guess what arbitrary thing I am thinking of next?” then the puzzle is broken. Sometimes, these are fixable. I was very impressed with the way that Nordhbane fixed his café puzzle (http://www.steamgifts.com/giveaway/gchWY/velocityultra). In the first incarnation of the puzzle, there were just a string of 13 x 8-bit binary words. The first three were designed to be converted via ASCII into characters. The other 80 bits needed to be interpreted in a different way. This was not sufficiently clued, and lots of people got stuck. Nordhbane, rewrote the puzzle, splitting off the first 24 bits onto a new line, and adding some very clever flavour-text designed to ensure that the method was well and elegantly clued, with more than one way into the solution. In the puzzles first incarnation, there was a bit too much of the “What am I thinking?” By the second version, the puzzle was very elegant. Now, if you do not solve it, you can legitimately facepalm when the solution is revealed. As a puzzle writer, I am always looking for that “Of course, Why didn’t I see that” moment in designing my puzzles.

In writing some of the sample puzzles, below, I will sometimes get myself into a situation where the puzzle has become “Now what am I thinking?” These are always situations where I need to step back and add additional cluing or just abandon my original idea.

Also as an aside note that everything provided is part of the puzzle. The title (if there is one) any flavour-text, and everything else provided are part of the puzzle. Even things that seem like errors could be important

This golden rule is even more important when writing puzzles for SG as opposed to other media, because what we are encoding is fundamentally meaningless. If you are coding an SG giveaway code of (say) A2bgY. That does not mean anything. So if you want to bold exactly 5 letters in your message to unambiguously identify the 5 characters, then that is fine, but if you are doing anything more involved, your answer phrase must use enough English to confirm that the solver is on the right track. For example, you could code “Congratulations, now your giveaway code is A2bgY”. That would be a lot fairer.

The goal of the puzzle-writer

A common misconception is that when you encounter a clever puzzle, a solver should feel that the puzzle-constructor is very smart. This is completely wrong. The goal of the puzzle-constructor is to make sure that the SOLVER feels very smart. I do not always succeed in my puzzles (this board is littered with noble failures), but when I do, the solvers feel pleased with themselves for solving it, When I succeed really well, the solvers struggle to solve it, but when they do, they think "I should have seen that sooner" and those who did not solve it, when they see the solution, should think "of course, how did I miss that." I have had a few puzzles like that, and its great as a constructor to achieve it.

The Field of Knowledge puzzle:

Sometimes, when I am reading, or working on a project, a puzzle idea will spring almost full-blown into my head, because I think “Hey, THAT would make for a cool puzzle. For example am sure that there is a really good puzzle in Complex Archaeological Statigraphy (which I learned about because I am related to an Archaeologist). It basically would be setting up a logic problem in terms that most readers are unfamiliar with. Someday I will write this puzzle, but it will take a vast quantity of research time to produce it.

Consider the following puzzle that was written by Jeremy White:
Family Affair
Maddy loves studying her family's genealogy using sites that search particular databases, such as Rootsweb, FamilySearch, book databases, and various Censuses. But not the ones you have to pay for.

• Mr. NameA is surprisingly missing from an 1899 family photo featuring Mr. and Mrs. Preston Crawford and their children.
• In the 1930s, genealogist Mr. Frederick Todd was considered the expert on the NameB family.
• NameC, grandchild of Samuel (b. 1716) and Elizabeth (b. 1714) Goodwin is remembered with a term we would consider an insult today.
• Living near James Haney (b. 1804) in Muscoda, Wisconsin in 1900 was a man whose family appears to have changed their surname to be NameD in English.
• In 1823, NameE Corrigal's marriage in the custom of the country was recognized by the Anglican church at the Red River Settlement in Manitoba.
• Mrs. Cole's memorial notice was published in a war-time edition of the Democrat Messenger on the 27th. Her maiden name was NameF.
• NameG is the family nickname of the man who married Nicolas Milleret's (b. 1690) widow.
• Mr. John Allen of Leicestershire married Miss Anne NameH in 1720.
  We are looking for the individual who was a descendant of Benjamin NameA, a descendant of Thomas NameB, and not a descendant of Mrs. NameC Willets NameD. This person was also a first cousin of NameE NameF's daughter-in-law, and died the same time of year as NameG Tarrant NameH. What was this person's first name?

Wow, what a task. It’s time for some genealogical research! To keep things easy, all of this information can be found using free internet resources. To keep things hard, the bare minimum amount of information needed is provided, and various different web resources are needed to be used. To solve this lots of standard genealogical research tricks may need to be used – searching using approximate dates or regions (because they might be off a bit, or a different term might have been used.), many alternate spellings of names, dropping information such as middle names, changing of date formatting, looking up definitions of terms. Web searching likes precision, and genealogical records are full of imprecision.

This sort of “puzzle” often lacks an AHA moment, but can be an interesting task nonetheless. There is often a sense of satisfaction that you get when you are able to pull something like this together.

Many logic puzzles, such as Sudoku also would fall into this class of problem, where there is a task that you need to solve, but there are only small AHA’s along the way.

Puzzles where [something] can produce a letter encoding

In some ways similar to the field of knowledge puzzle, this relies on the fact that certain keys indicate letters. So, for example, you think of a very clever way to cue Noble Gases (say), and then you sit down with a crossword solver to try to figure out which words can be made up of some or all of (He, Ne, Ar, Kr, Xe, Rn…). Ideally something longer than “HEAR”, but nothing immediately springs to mind. This is a serious disadvantage of a clever signalling method, that it inevitably supports only a very limited number of answer words.

There are many possible methods of signalling an alphabet. For example Sue Grafton wrote a series of detective stories featuring Kinsey Millhone with alphabetic titles: A is for Alibi, B is for Burglar, C is for Corpse and so on. We could theoretically signal some of these words, and as a result, clue the alphabet.

Or consider this puzzle by Noah Snyder and David Speyer which I have adapted for use here (and retitled):
Gorey Ends
Jane Austen
Robert Skaf
The Lost Prince
Tupac Shakur
Dante's Lucifer
Mary Jo Kopechne
Milton Krest (in the film)
Andres Escobar
Sergio Zardini
Joan of Arc
The Bedroom Killer
Oetzi, eventually
Arthur M. Free
Joe Wilson's neighbor's little girl (According to Henry Lawson)

Now where would we start with a puzzle like this? Some of these names have many associations, Jane Austen for example, but others are known for much less. Mary Jo Kopechne was the passenger who died in a single car accident at Chappaquiddick Island, when US Senator Ted Kennedy drove off a bridge and into the water. Oetzi was the Neolithic “Iceman”. Milton Krest is a villain in the James Bond film Live and Let Die, who suffers quite a spectacular death, and Joe Wilson’s neighbour’s little girl shows up for one sentence in Henry Lawson’s short story “Brighten’s Sister-in-Law: “I was scared now. I remembered a neighbour of ours had a little girl die (she swallowed a pin), and when she was going she said—‘Take the blankets off me, muvver—I’m dying.’And I couldn’t get that out of my head.” That is it. Seemingly the only thing notable about her is the way she died. And now we have the possibility of an AHA moment. Edward Gorey (of the title) was an American writer and artist famous for his macabre Victorian scenes. Amongst his works is a book called the Gashlycrumb Tinies, which is reproduced here: (https://www.brainpickings.org/2011/01/19/edward-gorey-the-gashlycrumb-tinies/) which describes and illustrates some unfortunate deaths (A is for AMY who fell down the stairs, B is for BASIL assaulted by bears, C is for CLARA who wasted away, D is for DESMOND thrown out of a sleigh). And with this we can solve the rest of the puzzle, by relating each of these people to the way that they died:

Jane Austen → C is for Clara who wasted away
Robert Skaf → A is for Amy who fell down the stairs
The Lost Prince → S is for Susan who perished of fits
Tupac Shakur → H is for Hector done in by a thug
Dante's Lucifer → W is for Winnie embedded in ice
Mary Jo Kopechne → I is for Ida who drowned in a lake
Milton Krest (in the film) → T is for Titus who flew into bits
Andres Escobar → H is for Hector done in by a thug
Sergio Zardini → D is for Desmond thrown out of a sleigh
Joan of Arc → R is for Rhoda consumed by a fire
The Bedroom Killer → A is for AMY who fell down the stairs
Oetzi, eventually → W is for Winnie embedded in ice
Arthur M. Free → A is for AMY who fell down the stairs
Joe Wilson's neighbor's little girl (According to Henry Lawson) → L is for Leo who swallowed some tacks
Yielding the answer CASH WITHDRAWAL.

Personally, I think that this puzzle, which worked successfully in its original context, would not work here. I just do not think that you can expect solvers to know of the existence of the Gashlycrumb Tinies, and without that key fact, the puzzle is unsolvable. It becomes the epitome of “What obscure book might I be thinking about”. That being said, the puzzle is easily fixable. You simply have to clue the book much more explicitly. For example, the title could be made clearer. If it was “The Gashlycrumb endings”, that would at least force an internet search.

Always keep on the lookout for what could clue alphabets. Musical frequencies can clue notes A through G. UK Ring sizes run A-Z, though I have never seen how to use that fact intelligently in a puzzle.

Start simple and obfuscate.

We start with the answer phrase “The Giveaway Code is XxXxX” and now we have to hide it. So what do we do?

We could translate it into an obscure language, but unless this is clearly motivated in the puzzle, it’s just crappy and worse un-fun. Yes, I accept that if the puzzle is supposed to be set in the Latvian embassy, then it is legal to write the puzzle in Latvian, but I find languages a cheap way to obfuscate, except in the context of a languages or linguistics puzzle.

We could set-up each letter separately in T H E G I V E A W A Y C O D E I S....
This could be as simple as a word list.

TREE
HOUSE
EAR
GIRAFFE
INSECT
VENUS
EGG
APPLE
WATER
Etc.

Now we have multiple ways of coding it.

Straight definitions

1. a plant having a permanently woody main stem or trunk
2. a building in which people live
   etc.

Fair, but fairly boring

Cryptic Crossword style definitions

(hmmm…. Writing cryptic clues is not my strength, why did I decide to write this guide again?):

1. Peter loses his head about woody plant
2. Residence of us around garden implement

Meh….

Clue each word with some form of series or grouping

Let’s say for each of our words, I could produce a string of images.

1. A long cigarette ash, Warrick Davis in character as Willow from the 1988 movie, US President Andrew Jackson, who was known as Old Hickory. The actress Thora Birch
2. Hugh Laurie in character as House M. D., a poker hand showing a full house from, the US White House, The symbols of Gryffindor, Slytherin, Hufflepuff, and Ravenclaw
   Etc.

Ok, this is getting slightly more interesting. But as a puzzle writer, I still don’t like it much. Let’s say I can identify Ash, Willow, Old Hickory and Thora Birch in the first example, how is that clearly “TREE” as opposed to say “WOOD”? Anytime there are lots of possible valid answers and the puzzle becomes “guess which synonym I am thinking of?” then I violate the golden rule and I need to reject the puzzle on arbitrariness grounds. This is fixable in this case. There could be a scrambled pile of additional pictures of an iconic tree, an iconic house, etc. but it is getting clunky. Particularly so as the things we are cuing (TREE, HOUSE, etc.) are not meaningful thematically.

Theme your word list

Instead of Tree, House, Ear, Giraffe, Insect, Venus, Egg, Apple, Water, etc. We could make our list out of operas:
TOSCA
HÄNSEL UND GRETEL
EUGENE ONEGIN
GÖTTERDÄMMERUNG
IOLANTHE
VENUS AND ADONIS
EINSTEIN ON THE BEACH
AIDA
WILLIAM TELL
ARIADNE AUF NAXOS

To cue this now we could include plot summaries of the operas. Or simply concatenate the first line of each overture into a single longer music track. That puzzle would be fair if near impossible. Nobody on this board will solve it, but it would be a fair puzzle.

Vary the theming

Consider a different theme such as:

University of Tennessee Volunteers
University of Hawaii Warriors
And then a set of definitions for volunteers, warriors, etc.

Or even

TERRAPINS
HUSKIES
EAGLES

Which could then be clued simply as:

MARYLAND
WASHINGTON (or CONNECTICUT)
EASTERN MICHIGAN
Etc.

Depending on the flavour-text this might work, or it might just be too arbitrary. I do not live in the US, and US University sport may just be too obscure for a global audience like that at SG.

Indexes

If we are allowed indexes, then our word choice becomes a lot less constrained. Consider for example:
Green, White, Red (2)
Blue, Yellow, Red (2)
Green, White, Orange (3)
Black, Yellow, Red (4)
Green, Yellow, Red (4)
Blue, Yellow, Green With 3 Green Diamonds (6)
Red, White, Red, (2)
Red, Yellow, Green (6)
Etc.

Now we could recognise these as tricolor flags and get:
Italy (2)
Chad (2)
Ireland (3)
Belgium (4)
Mali (4)
Saint Vincent and the Grenadines (6)
Peru (2)
Guinea (6)
Etc,
Or using the number as an index into the answer we get THE GIVEA…

Or do it with numbers:
TWELVE MINUS TEN (ONE)
SQUARE ROOT OF NINE (TWO)
TEN HALVED (FOUR)
FIVE PLUS THREE (THREE)
THREE TIMES THREE (TWO)
TWO PLUS THREE (THREE)
THREE MINUS TWO (THREE)
Etc.

Or
TWO (1)
THREE (2)
FIVE (4)
EIGHT(3)
NINE(2)
FIVE(3)
ONE (3)
Or THE GIVE…

Could we make it more Obscure:
TWELVE MINUS TEN.
SQUARE ROOT OF NINE..
TEN HALVED….
FIVE PLUS THREE…
THREE TIMES THREE..
TWO PLUS THREE…
THREE MINUS TWO…

I still think that this is fair. The random number of periods after the operations is certainly strange, and when puzzle solving, embrace the strange. It is probably there for a reason.

For what it is worth, I do not think that this would be fair
12-10.
2+1..
10/2….
Etc. There is nothing to clue that you are to use the spelled words for the name of the digits rather than the digits themselves, so this strikes me as a “What arbitrary thing and I thinking about?” puzzle and it fails under the golden rule.

Sort

Another nice way to obfuscate the puzzle is to put the relevant information in the wrong order. With this mechanism, you really need an unambiguous sort order for the things in your list. In my puzzle *Let Me Make Myself Perfectly Clear” (http://www.steamgifts.com/discussion/0Dla6/mikalyes-9th-2016-puzzle-let-me-make-myself-perfectly-clear-ended-solution-posted), I set up a series of trivia question whose answers have a clear and unambiguous ordering. After solving the questions, you then need to reorder them.

This only works with things that have a clear ordering

• each day I tell Boris Levit, “I customarily have pancakes for breakfast”
• George’s luckiest number was the number seven.
• he found the Convexo Dust Buster to be the finest handheld vacuum cleaner
• Today in Oregon, in Eugene, Sister Mary Francis led the choir

Here you need to notice that each sentence has an embedded book of the bible (Leviticus, Numbers, Exodus, Genesis), putting them in order we get

• Today in Oregon, in Eugene, Sister Mary Francis led the choir
• he found the Convexo Dust Buster to be the finest handheld vacuum cleaner
• each day I tell Boris Levit, “I customarily have the pancakes for breakfast”
• George’s luckiest number was the number seven.
  Or the start of THE G…[IVEAWAY]..
  P.S. I warned you that the puzzles I would create on the fly as I wrote this thing would suck.

ISIS Puzzles

We can combine sorts with indexes and get something very flexible. This is the idea behind the ISIS puzzle
Identify (recognize the film quotes, the song lyrics, the game characters, whatever)
Sort (put them into the unambiguous ordering)
Index (Pick the correct element of your answers)
Solve (Pretty self-explanatory really)

Encode your message.

Once, desperate for a puzzle to fill a slot at the last minute, I took the message, wrote it in Braille. Took each line of the Braille which could either be ○○, or ○●, or ●○, or ●● and treated that as O, N, A, or I by treating them as Morse code. Converted those into semaphore, which I expressed as times, and then wrote extensive flavour-text to hint at what I had done. I viewed this puzzle as the embodiment of concept of “filler”, but I needed one more puzzle to make a multi-stage multi-puzzle event work, and I was right up against a hard deadline. I hated this puzzle. Many of the solvers liked it. I was both saddened and a little depressed that this worked, but code works very well.
Make sure that you know your standard encodings. Apart from the one’s listed in Zelghadis’ guide, make sure that you can recognise Braille, Sign Language Fingerspellings, the Pigpen cipher, Roman Numerals, Semaphore and the (underused on SG) Nautical signal flags.

We can even provide our own codes

Back to our need to encode “The Giveaway Code is XxXxX”, we could define a simple letter substitution “SLF RIE…” where T=S, H=L, E=F etc. and then clue famous authors and/or poets who write in English
T. S. Eliot
H. L. Mencken
E. B. White
G. K. Chesterton
I. F. Stone
V. S. Naipaul
[E has been defined above]
A. A. Milne
W. H. Auden
[A has been defined above]

This sort of coding sometimes includes a fair amount of research to get a grid of assignments. That is to say, how can I find 26 names that give me a sufficient universe for my conversion, or do I just not care. Also note that I screwed up the above set. When I get an S, does that convert to a T as in T. S. Eliot, or to a V as in V. S. Naipaul. We would need to fix that, perhaps by replacing T. S. Eliot with T. E. Laurence or T. J. Binyon.

This sort of coding is the basis of a lot of programming code problems as well.

Metapuzzles

Sometimes an event has a final round, which requires you to combine the answers that you have worked out in previous puzzles into a final solution. This is known as a metapuzzle, a puzzle that unites several puzzles that feed into it. There are basically two types of Metapuzzles, "pure" metapuzzles and "compound" or "shell" metapuzzles. The "pure" meta uses a pattern found solely in the answer words. The "shell" meta includes separate, additional puzzle content

A pure meta consists simply of a list of words. The words have some property that would allow you to identify another word that can solve the puzzle. This word can be based on either the meaning or some property of the words, such as letter patterns. For example, suppose you had solved eight puzzles and gotten the answers: BLACK, HAMMER, FROST, LUMBER, POT, KNIFE, and UNION. The answer to the metapuzzle based on those answers might be "JACK", which can be paired with each of the other words to make a longer word or common compound (blackjack, jackhammer, jack frost, lumberjack, jackpot, jackknife, and union jack). This is a metapuzzle. Note that sometimes you do not need to have all of the previous answers in order to solve a metapuzzle. Here you could reasonably guess the answer with four of the five answers, and indeed possibly as few as three.

Consider the following meta that appears in the 2011 MIT mystery hunt. You know that the theme of this puzzle is the Mario Bros. series of video games. You have 8 of the 10 previous answers. Can you spot the meta solution?
???
METALLURGICAL
UPROARIOUS
PLUNGING
SUMMARIZE
BLUEGILL
IMPRESARIO
SUPERFLUITIES
???
SPECTRAL RIGIDITY

Hmmmmm....Here what matters is the arrangement of letters. You might notice that each of the words ALMOST contains the strings "MARIO" or "LUIGI" alternating. I say almost because each have one letter wrong. Let's capitalise these.

???
metalLURGIcal
uprOARIOus
pLUNGIng
sumMARIZing
bLUEGIll
impreSARIO
superfLUITIes
???
spectraLRIGIdity

Let's now consider the wrong letters. We get ?RONZEST?R
The answer to the metapuzzle is BRONZESTAR.

In my puzzles, A LIST OF WORDS, was basically a pure metapuzzle.

In a constructed or compound meta, there is an additional thing provided. This can be a fill-in grid, a diagram, or something else. In jeffhowes Food pyramid puzzle, you had to enter the answers from an ITSTOOHARD quiz into a crossword grid to extract the next step. This is really a shell meta

Backsolving

Backsolving arises when you can guess a solution to an existing missing puzzle based on the structure of the metapuzzle. Consider the first puzzle I used to explain metas. Suppose you had BLACK, HAMMER, FROST, [unknown], POT, KNIFE, and UNION and you have solved the meta to be JACK. Now you know that the [unknown] solution has to be a word that is immediately before or after jack. You might guess CRACKER, JUMPING, FLAP, RABBIT, RUSSELL, APPLE, until you finally guess LUMBER. If you can correctly guess a solution from an understanding of the Metapuzzle without actually figuring out the puzzle, you are backsolving.

Consider the second more constrained puzzle. We know that the missing first word must contain "MARIO", but with a B replacing one of the other letters, so it contains either BARIO, MBRIO, MABIO, MARBO, or MARIB. You can plug these strings into some of the word search tools I list below in references and get very few options. There is the capital of Suriname, PARAMARIBO, there is BARBARIOUS, there is OBAMA BIOGRAPHY. there is the sound engineer who created the speaker company AMAR BOSE. Not really many viable options. We can reasonably backsolve this puzzle.

Puzzle hunts

Occasionally a larger event will run on SG (jeff howe has run a few), which may stream together multiple puzzles, possibly with multiple metapuzzles. This is a puzzle hunt. Some can be gigantic the MIT Mystery hunt can run to 170 puzzles, but there are smaller ones with only 10 puzzles or so. There is a list of puzzlehunts on the Wikipedia page, many of which are online.

Reference Materials

When I set out to write the flag id puzzle above, how was I able to be sure that I distinguished correctly between Guinea (red, yellow, green) and Mali (green, yellow, red). Well I looked them up. There are a number of flag ID sites out there, of which my favourites are http://www.flagid.org/ and http://www.onlinestores.com/flagdetective/ and http://www.photius.com/flags/flag_identifier.html

For crossword clues, you might try: One Across http://www.oneacross.com/crosswords/ or crossword tracker http://crosswordtracker.com/ or Crossword Heaven http://www.crosswordheaven.com/

To solve a cryptogram, or any other substitution cipher, I recommend Decrypto or quipquip http://www.quipqiup.com/

There is also a value in pattern matchers. I use OneLook (http://www.onelook.com/) and Design215 (http://design215.com/toolbox/wordpattern.php) and quinapalus (http://www.quinapalus.com/cgi-bin/match). Quinapalus has a variety of interesting tools on their site including the incredibly cool Crux (http://www.quinapalus.com/crux.html), and a playfair cipher breaker (http://www.quinapalus.com/cgi-bin/playfair)

Actually I use offline tools for these as well, my favourite is TEA (http://www.crosswordman.com/tea.html). The best online version of this is https://nutrimatic.org/ For those Hitchhiker's Guide fans, you will know that a Nutrimatic produces something that is "almost, but not quite, entirely unlike tea."

When doing lookups, there is the need for a good word list. I recommend Moby (http://icon.shef.ac.uk/Moby/) which includes codings for thesauri, for Shakespeare, for phonetic alphabets and multiple languages.

Other reference tools include ASCII and EBCDIC tables, dialing codes and HTML (http://www.lookuptables.com/) Braille, Caesar ciphers, Beale Ciphers, and Pigpen Ciphers and Morse Code tables all are available on Wikipedia. Nautical flags can be found here: http://www.pem.org/sites/archives/guides/signals.htm#itc. And solving phone spelling problems can be found here: http://www.phonespell.org/ albeit with US phones which do not have Q’s or Z’s (but that would still solve most of my Phoney numbers puzzle). Semaphore is here http://www.anbg.gov.au/flags/semaphore.html or here http://inter.scoutnet.org/semaphore/semaphore.html and Unicode is here: http://unicode.org/

Other useful reference tools include Web Elements (http://www.webelements.com/) and odd things like What’s special about that number (http://www2.stetson.edu/~efriedma/numbers.html) rhyming dictionary (http://www.rhymezone.com/) and the encyclopedia of integer sequences (https://oeis.org/). There is even a website which asks What's Special about this Number? http://www2.stetson.edu/~efriedma/numbers.html

There are times to use hashing algorithsm, and you can find a bunch of them here (https://hash.darkbyte.ru/~dHlwZSBoZXJl)
To translate between binary, ascii, hex, base64, etc. you can use x-late here (https://paulschou.com/tools/xlate/)
Find a truckload of code tools here (http://www.dcode.fr/tools-list)
Recognise that some of the text may be obfuscated to make it harder to read (http://obfuscator.uo1.net/)
And do find a large number of text manipulations here (http://qaz.wtf/u/convert.cgi?text=the+quick+brown+fox)

Now Zelghadis does a much better job than I do of covering all of the nasty things that can be done with image handling, but it is worth mentioning a few online image manifpulation tools. For viewing metadata, I usually use (http://metapicz.com) or (https://onlineexifviewer.com/) For a hex-editor, you can try (http://en.webhex.net/) or (https://www.onlinehexeditor.com/)

Finally if you use an iPhone you definitely need the free Puzzle Sidekick (https://itunes.apple.com/us/app/puzzle-sidekick/id678644111?mt=8&ign-mpt=uo%3D4)

Hope this helps,
Mikalye