Colemak, Dvorak or QWERTY: Which One Better?
Colemak, Dvorak or QWERTY: see wetin each layout optimise, why speed evidence no strong, and why switching fit mean weeks of slow work.
Di short answer
Di Colemak vs Dvorak vs QWERTY question get one boring answer for most people: stay for QWERTY, because you go pay the cost with weeks of slow work, while the benefit small and hard to measure. If you really wan move, Colemak na the cheaper move. E leave 13 out of the 30 letter and punctuation keys exactly where QWERTY get dem, and e keep the copy and paste keys for the same place. Dvorak na the move wey go disturb things pass, but e get the widest support, because every mainstream operating system don ship am for decades. Neither layout go make you type faster by itself. Wetin dem change na how far your fingers dey travel, and how e dey feel after long day.
Dis page na the decision. If you don already decide and you want the commands, go the Colemak setup guide for Linux.
Wetin each layout design to do
QWERTY come from the Sholes and Glidden typewriter of the 1870s, and na mechanical problem e dey solve. For that machine, every key dey swing one metal typebar up from one basket to strike the paper. Two bars wey dey close to each other fit clash and jam if person strike dem one after another quickly. To spread common letter pairs apart reduce the jams. The popular story say QWERTY design to slow typists down na distorted version of wetin happen. Historians don also trace some parts of the arrangement to telegraph operators wey dey transcribe Morse code. The important thing for you na this: 1878 hardware freeze the layout, then decades of training lock am in place. Nobody tune any part of am for keyboard wey no get moving parts.
Dvorak na the alternative from August Dvorak and William Dealey wey dem publish for 1936. You fit see the design rule from the layout itself. All five vowels dey the left home row as AOEUI, while the most common English consonants dey the right home row as DHTNS. Common words then dey alternate between both hands, and the strongest fingers dey do most of the work. E also disrupt the three layouts pass by far, because na only two keys, A and M, still dey their QWERTY positions.
Shai Coleman publish Colemak for 2006, and the design constraint na the switching cost itself. The home row read A R S T D for the left side and H N E I O for the right side. This one puts the most common English letters under your fingers, while everything else moves as little as possible. The bottom row keep Z X C V B for the same places wey QWERTY get dem. Colemak also remap Caps Lock to Backspace. This na separate habit wey you fit adopt for any layout or ignore completely.
Colemak vs Dvorak vs QWERTY: how many keys really dey move
Count the three letter rows, wey get 26 letters and 4 punctuation keys, then compare every layout with QWERTY position by position. The shortcut column dey count how many of Z, X, C and V remain for their QWERTY positions, because na those four dey carry undo, cut, copy and paste for most systems.
The data behind this chart
[
{
"label": "QWERTY",
"keys_moved_from_qwerty": 0,
"shortcut_keys_kept": 4
},
{
"label": "Colemak",
"keys_moved_from_qwerty": 17,
"shortcut_keys_kept": 4
},
{
"label": "Dvorak",
"keys_moved_from_qwerty": 28,
"shortcut_keys_kept": 0
}
]Colemak dey move 17 keys. Dvorak dey move 28. Na this gap be the whole practical difference between the two. Every key wey move na one habit you need unlearn while your hand still dey reach the old position, so the count na rough measure of how long the painful part go last. E no talk anything about which arrangement better.
Editing shortcuts dey stay where you expect?
Colemak keep Z, X, C and V for their QWERTY positions, so Ctrl+Z, Ctrl+X, Ctrl+C and Ctrl+V still dey under the same fingers for the same hand. Dvorak keep 0 of dem. For Dvorak, letter C dey where QWERTY get I, V dey for QWERTY full stop key, X dey under QWERTY B, and Z dey for QWERTY slash key. Copy and paste become reach wey need both hands.
This matter pass for terminal than for word processor. Ctrl+C to interrupt command wey dey run and Ctrl+D to close standard input na things wey server operator dey press hundreds of times every day. For Dvorak, both move go right hand: C go QWERTY I key, and D go QWERTY H key. macOS ship variant wey change back to QWERTY while you hold Command key. Linux no get standard equivalent for Ctrl key, so for Dvorak the shortcuts move and dem stay moved.
Vim and less move cursor with h, j, k and l, and dem choose these letters because dem dey under right hand for QWERTY. Both alternative layouts break this arrangement. For Colemak, letters wey dey printed on physical QWERTY h, j, k and l keys na h, n, e and i, and the usual fix na to remap movement to those four. For Dvorak, the same four physical keys type d, h, t and n. Either fix na a few lines for your vimrc, and e stop to help immediately you dey use another person machine with stock config.
Wey evidence for typing speed no strong
Part of the argument na arithmetic, and nobody dey dispute am. If you get one layout and one body of text, you fit calculate the total distance wey fingers travel and how often one finger need type two letters one after another. Both numbers dey lower for Dvorak and Colemak than for QWERTY when na ordinary English prose, and layout analysers agree on the direction. But the results depend on the text wey dem use calculate am. Your source code, your language, and your variable names no be that kind of text.
The part wey people dey dispute na whether the arithmetic really turn into speed. The strongest claim for Dvorak come from one United States Navy study during the Second World War. The study report large gains after retraining. But August Dvorak supervise the study himself, so e no be independent evidence. For 1956, Earle Strong run one controlled retraining study for the General Services Administration. E find say Dvorak retraining no beat giving QWERTY typists the same extra practice. For 1990, economists Stan Liebowitz and Stephen Margolis gather these criticisms for one paper wey dem call The Fable of the Keys. People still dey argue about that paper, because the real subject na market lock in, no be keyboards. After ninety years, no large and well-controlled trial still show say alternative layout make people type faster.
There dey one simpler reason to handle speed claims carefully. People don type well pass 150 words per minute on QWERTY. If you dey type at 60 or 70, na not the layout dey limit you.
Comfort and injury na harder questions, and evidence for dem even weaker. People dey report say their hands hurt less after dem switch, and that one worth taking seriously. But person wey switch usually change several things at once. New keyboard, different desk height, more breaks, and several weeks of deliberately slow typing fit all happen for the same month. Any one of dem fit be the cause. If pain na your reason, fix your posture and workload first. Those ones get better evidence behind dem, and dem no cost anything.
Wetin switching really cost
Expect say for the first few days, you go dey type like beginner, around 15 words per minute. E fit take two to four weeks before the layout stop to take all your attention. To reach your old speed again commonly dey take one to three months of daily use. Na figures wey people wey do am report, no be study result, so treat dem as range, no be promise.
The part wey dey surprise people na interference. Your hands no dey learn "Colemak". Dem dey learn one destination for each key. If you use two layouts for alternate hours, both mappings go remain weak. To switch back to QWERTY for urgent task, then switch again, na the slowest way to learn. Two approaches dey work. Switch once and accept the bad week. Or keep a hard split, where one physical keyboard always use one layout. That way, the context go tell your hands which mapping to load.
Most people wey give up dey do am during that dip. Dem give up because the week busy, no be because the layout no fit dem. That one make timing the most important decision for the whole switch. Start during quiet week. No start the week before deadline, and no start the week wey you dey on call.
Your QWERTY skill usually go survive. Most people wey switch report say e dey return within few days of use, but e slower than before and mistakes dey happen at first. E good make you know this, because na losing the skill people dey fear pass. But na that part turn out to be least true.
Client or server: remap go belong for where?
Almost every time, na client. Keyboard layout dey apply for the machine wey keyboard attach to, and nowhere else. When you press key, your local system dey turn that key position into character, then the SSH session wey dey carry your keystrokes dey send the character go server as byte. Server no dey see key press at all. So if you run localectl set-keymap dvorak for remote box, e no go change how your typing dey arrive through SSH. Workstation wey set to Colemak go type Colemak into every host wey you log into, without any configuration for those hosts.
The server own keymap matter for two places: keyboard wey physically attach to the machine, and console wey your provider give you through browser or serial link. For VPS, na rescue console this mean. Na exactly the situation wey you fit least afford to experiment.
localectl status
localectl list-keymaps | grep -i dvoraklocalectl status dey print both settings: VC Keymap for text console and X11 Layout for graphical session. list-keymaps dey print the console keymap names wey your system actually get installed. This check stop you from setting name wey no exist. Colemak no dey always available as console keymap even when e dey available for graphical session. So run that grep for your own layout before you assume say e dey there.
For desktop side, setxkbmap -layout us -variant colemak dey change the layout of running X session, while localectl set-x11-keymap us pc105 colemak dey make the choice survive reboot. Under Wayland, compositor dey control the keymap. So setxkbmap no be the correct tool there. The change belong for your desktop input settings or inside localectl. The complete procedure, including wetin to do with Caps Lock, dey inside the Colemak for Linux walkthrough.
Why keyboard firmware fit sidestep the whole problem
If you dey log into plenty machines, put the layout for inside keyboard instead of operating systems. Programmable keyboard dey store its own mapping and send code for the letter wey you want, so every host go see ordinary keyboard dey type ordinary letters. Nothing for any host need configuration, and nothing fit forget during rebuild. This one still cover machines wey you no fit configure at all: colleague laptop, UEFI setup screen, rescue image, or host wey you no get permission to change system settings. For anybody wey must keep plenty Linux servers in order, this na the difference between one purchase and configuration item for every box.
The common way na keyboard wey dey run QMK, or its wireless relative ZMK. Both na open firmware wey you flash by yourself. VIA and Vial na graphical editors wey fit write change without rebuild. Plenty mainstream keyboards too dey ship vendor software wey stores remap for inside keyboard. The limits dey clear. Laptop built-in keyboard usually no fit do this, so you get your layout when external keyboard dey plugged in and QWERTY when e no dey, and that one na another kind of interference.
Rescue console na real operational risk
The failure dey happen like this. Server no come back after reboot, you open provider web console, then you need type root password and some filesystem commands with layout wey you no dey type well again. Two things fit go wrong at the same time. Your muscle memory don now follow the new layout, but the console fit no follow am.
Browser consoles no dey send keystrokes the same way. Some send the character wey your local system produce, so your client-side remap follow you. Others send key position and assume say US QWERTY map dey for the other side. For that case, your Colemak keyboard go type QWERTY letters inside the console. You no fit know which type you get just by reading control panel. Test am while the machine still healthy: open console, log in, type one sentence, then check wetin appear. Do am now, no be during outage.
Then remove the parts wey you fit remove. Log in with keys so ordinary access no go ever ask you to type password, and treat key management across your servers as part of recovery plan, no be just convenience. Keep rescue password for place wey you fit paste from, because most browser consoles get send text or paste function. Also keep some way to find QWERTY letters by looking keyboard, because to read layout chart from your phone at 2 in the morning dey slow.
Encrypted disks add one more trap. Passphrase prompt for boot dey run from initial ram disk, before your normal keymap load, so e use any map wey dey built into that image. For Debian and Ubuntu, na built-in US QWERTY map e dey use by default. The setting dey for KEYMAP= inside /etc/initramfs-tools/initramfs.conf, and e ship as n. Set am to y, run sudo update-initramfs -u, then reboot and confirm say passphrase prompt dey accept your layout before you trust am. Machine wey no fit unlock na machine wey you go need rebuild.
Colemak-DH and the other options
Choice no be only three options. Colemak-DH, wey dem dey also call Mod-DH, na the most common improvement for Colemak. For standard Colemak, letters D and H dey the middle columns of the home row. Index fingers reach dem by stretching sideways instead of curling. Mod-DH move these two letters go the bottom row under the index finger. This give home row wey be A R S T G for left side and M N E I O for right side. E get two versions: one for the usual row-staggered keyboards, and one for ortholinear and column-staggered boards. You need choose the one wey match your hardware.
Apart from that, Workman, Norman, Canary, Graphite and plenty other layouts dey. Layout optimisers produce many of dem. Programmer Dvorak rearrange the number row and symbols for people wey dey type punctuation pass normal prose. Other languages get their own layouts too, like bépo for French and Neo for German. The practical thing to check na availability. Dvorak dey available almost everywhere, including phone keyboards and console keymaps. Colemak dey included with the X keyboard configuration data wey every Linux desktop dey read. Colemak-DH enter that data more recently. So, for older distribution, you may still need install the layout files from the project. Community layout na file wey you must carry go every machine, or bake inside your keyboard firmware.
So, which one you suppose choose?
If nothing dey wrong, remain with QWERTY. The switch fit take weeks, and evidence say e go make you faster no strong. So, small irritation with your current layout no be strong enough reason by itself.
Choose Colemak if you want rearrangement wey go cost you less. Thirteen keys remain for the same place, editing shortcuts remain for the same place, and na one-line change for any Linux desktop.
Choose Dvorak if you value setting your layout from menu wey already dey available for any machine and any operating system, and you accept to relearn the shortcut keys.
Choose Colemak-DH if you already dey buy programmable keyboard. The firmware remove the availability problem, so you fit as well use the improved version.
No matter wetin you choose, decide where the mapping go dey before you start. If you spend the day for full-time terminal workbench and you dey log into many machines, keyboard na the correct place for am. If you work on one laptop, operating system dey okay. Then plan the rescue path, and test am before you need am.
FAQ
Colemak or Dvorak really dey faster pass QWERTY?
No solid evidence dey show say either one go make you type faster. Wetin person fit measure na the calculation: for English writing, both layouts put more keystrokes for the home row and make fingers travel less than QWERTY. But no large, well-controlled trial don show say this one dey produce higher words per minute. Dvorak Navy study wey people dey quote most, Dvorak himself supervise am. Earle Strong 1956 study for General Services Administration find say no advantage remain after QWERTY typists get the same extra practice. People dey reach well past 150 words per minute with QWERTY, so if your speed na 70, the layout no be your limit.
How long e dey take to learn new keyboard layout?
Plan for some days of typing near 15 words per minute, two to four weeks before the layout stop taking all your attention, and one to three months to reach your former speed. These na reported ranges, no be study results. The bigger risk na interference: if you dey alternate between two layouts during the day, both go remain weak. So either switch once and accept the slow week, or keep strict split where one physical keyboard always mean one layout. Most people wey quit dey do am during that slow period because the week become busy, so start for one quiet week.
My keyboard layout dey follow me over SSH?
Yes, and the server no get control over am. The layout dey apply for the machine wey your keyboard connect to. Your local system turn the key press to character, then SSH send that character as byte, so the remote host no dey see key position. Setting keymap for the server no change anything for SSH users. Server own keymap only apply to keyboard wey connect directly, or console wey provider give you for browser or through serial link.
Wetin go happen to Ctrl+C and vim h j k l keys?
Colemak leave Z, X, C and V for their QWERTY positions, so standard editing shortcuts no move. Dvorak move all four: C land on QWERTY I key, V on QWERTY full stop, X under QWERTY B, and Z on QWERTY slash. Ctrl+D move to QWERTY H key, and this matter for shell. Cursor movement break for both layouts. Colemak users usually remap movement to h, n, e and i. These na the letters wey dey printed on physical QWERTY h, j, k and l keys. Dvorak users use d, h, t and n for the same reason.
How I fit keep rescue console usable after I switch layout?
Test am now, while the server still healthy. Open your provider console, log in, and type one sentence. Some consoles forward the character your client produce, while others forward the key position and assume US QWERTY for the other side. Log in with SSH keys so routine access no ever need typed password. Keep any rescue password for place wey you fit paste from, because most browser consoles get paste or send text function. If disk get encryption, boot passphrase prompt use keymap wey dey inside initial ram disk. For Debian and Ubuntu, that mean setting KEYMAP=y inside /etc/initramfs-tools/initramfs.conf, running sudo update-initramfs -u, then rebooting to confirm say the prompt accept your layout.