Max Schillinger <maxschillin...@web.de> writes: >> Ich finde dazu den Optimierungsansatz von KOY recht überzeugend – kann man >> irgendwo nachlesen, wie Bone optimiert wurde? > Bone ist im Internet tatsächlich unterrepräsentiert. Ich hab mal durch > den AdNW-Optimierer geschickt. Die Kennzahlen für bone auf der > AdNW-Homepage <http://www.adnw.de/uploads/Main/Grafiksammlung/boneD.pdf> > scheinen nur für deutsch vorhanden zu sein. Außerdem ist da nicht das > bone zu sehen, das in Linux integriert ist. Beim Linux-bone kommen mit > dem AdNW-Optimierer für deutsch+englisch diese Zahlen raus: > > bone 443.211 Gesamtaufwand 214.682 Lageaufwand links > rechts > 2.034 Kollisionen 1.982 Shift-Kollisionen ob > 14.0 13.6 > jduax phlmwß 63.058 Handwechsel 33.581 Shift-Handwechsel mi > 35.5 24.7 > ctieo bnrsgq 0.986 Ein-/Auswärts 32.462 Ein- oder auswärts un > 5.6 6.6 > fvüäö yz,.k 13.344 benachbart 4.749 Shift-benachbart sum > 55.1 44.9 > 7.0 12.2 10.6 25.3 --.- --.- 17.0 11.3 9.7 7.0 Sh > 2.2 1.9 > > Das kannst du mal mit der Tabelle hier vergleichen: > https://maximilian-schillinger.de/keyboard-layouts-neo-adnw-koy.html
Vorsicht beim Vergleich von Zahlen, da bone aus einem anderen Optimierer stammt als AdNW. Für die "eigenen" Ergebnisse wird ein Optimierer immer bessere Zahlen liefern, als für andere, weil die ganzen kleinen Entscheidungen und Kriterien da gleich sind, obwohl unklar ist, was eigentlich der genau beste Wert für Parameter ist. In meinem Optimierer in der aktuellen Version, die sich ein Stück von der von Bone wegentwickelt hat, sind Bone und AdNW etwa gleich gut. Das heißt, es ist wahrscheinlich, dass AdNW in realer Nutzung ein Stück glatter läuft, allerdings nicht auf den 10% Unterschied kommt, den der AdNW Optimierer nennt. Die Ergebnisse, die ich zur Zeit habe, sind $ ./check_neo.py -v Neo # Evolved Layout xvlcw khgfqyß uiaeo snrtd⇘ üöäpz bm,.j # 19.14649 x100 total penalty per letter # 70.07100 x10 billion total penalty compared to notime-noeffort # 7.351595 mean key position cost in file 1gramme.txt ( 53.80972458 ) # 41.33912 % finger repeats in file 2gramme.txt ( 199.5607999384172 ) # 12.23492 million keystrokes disbalance of the fingers ( 24.46984395255216 ) # 0.858586 % finger repeats top to bottom or vice versa ( 20.723734419867274 ) # 2.526418 % of trigrams have no handswitching (after direction change counted x 1 ) ( 11.175141922045984 ) # 13.91644 billion (rows²/dist)² to cross ( 180.91372047652894 ) # 0.008960 hand disbalance (4% shift is too much). Left: 0.49103950950814895 %, Right: 0.5089604904918511 % ( 0.2623438746694232 ) # 0.365973 badly positioned shortcut keys (weighted). # 22.29890 no handswitching after unbalancing key (weighted). # 19.45849 movement pattern cost (weighted). # 0.847645 asymmetric bigram cost (weighted). # 97.19161 irregularity cost (weighted). # 14.31105 manually assigned bigram penalty (weighted) # 14.28082 unbalancing key after neighboring finger (weighted) # 0.847645 asymmetrc bigrams (weighted) # 41.04022 inconsistent symmetries of similar keys (weighted) Qwertz for comparision # Evolved Layout qwert zuiopü+ asdfg hjklöä <yxcvb nm,.- # 40.07729 x100 total penalty per letter # 146.6721 x10 billion total penalty compared to notime-noeffort # 12.11769 mean key position cost in file 1gramme.txt ( 88.6950124 ) # 62.18356 % finger repeats in file 2gramme.txt ( 300.1853743760975 ) # 10.35744 million keystrokes disbalance of the fingers ( 20.71489352786094 ) # 16.17628 % finger repeats top to bottom or vice versa ( 390.4476026206619 ) # 6.237203 % of trigrams have no handswitching (after direction change counted x 1 ) ( 27.589113184590314 ) # 25.14135 billion (rows²/dist)² to cross ( 326.83764675635734 ) # 0.066360 hand disbalance (4% shift is too much). Left: 0.566360286279346 %, Right: 0.433639713720654 % ( 1.9428863456222933 ) # 0.731946 badly positioned shortcut keys (weighted). # 30.53207 no handswitching after unbalancing key (weighted). # 32.10186 movement pattern cost (weighted). # 0.807610 asymmetric bigram cost (weighted). # 116.3805 irregularity cost (weighted). # 8.088897 manually assigned bigram penalty (weighted) # 46.37644 unbalancing key after neighboring finger (weighted) # 0.807610 asymmetrc bigrams (weighted) # 75.28918 inconsistent symmetries of similar keys (weighted) And AdNW # Evolved Layout kuü.ä vgcljf´ hieao dtrnsß xyö,q bpwmz # 13.15343 x100 total penalty per letter # 48.13804 x10 billion total penalty compared to notime-noeffort # 7.701480 mean key position cost in file 1gramme.txt ( 56.37069156 ) # 11.20421 % finger repeats in file 2gramme.txt ( 54.08728726147864 ) # 12.07988 million keystrokes disbalance of the fingers ( 24.15977418430607 ) # 0.607672 % finger repeats top to bottom or vice versa ( 14.667411396945747 ) # 1.776137 % of trigrams have no handswitching (after direction change counted x 1 ) ( 7.856416105788094 ) # 8.807276 billion (rows²/dist)² to cross ( 114.49459044655136 ) # 0.037646 hand disbalance (4% shift is too much). Left: 0.4623539453386664 %, Right: 0.5376460546613336 % ( 1.1021954495518846 ) # 1.097919 badly positioned shortcut keys (weighted). # 17.29860 no handswitching after unbalancing key (weighted). # 37.87753 movement pattern cost (weighted). # 0.859672 asymmetric bigram cost (weighted). # 63.67743 irregularity cost (weighted). # 18.38554 manually assigned bigram penalty (weighted) # 16.34324 unbalancing key after neighboring finger (weighted) # 0.859672 asymmetrc bigrams (weighted) # 53.10210 inconsistent symmetries of similar keys (weighted) And CRY # Evolved Layout bmuaz kdflvjß criey ptsnh⇘ xäüoö wg,.q # 13.18032 x100 total penalty per letter # 48.23642 x10 billion total penalty compared to notime-noeffort # 7.589013 mean key position cost in file 1gramme.txt ( 55.5474968 ) # 14.03951 % finger repeats in file 2gramme.txt ( 67.77444941174628 ) # 11.68242 million keystrokes disbalance of the fingers ( 23.36485368610392 ) # 0.510983 % finger repeats top to bottom or vice versa ( 12.3336314588947 ) # 3.606038 % of trigrams have no handswitching (after direction change counted x 1 ) ( 15.950643910455666 ) # 8.795237 billion (rows²/dist)² to cross ( 114.3380911999914 ) # 0.027941 hand disbalance (4% shift is too much). Left: 0.5279417425876841 %, Right: 0.4720582574123159 % ( 0.8180740799998765 ) # 0.731946 badly positioned shortcut keys (weighted). # 14.07224 no handswitching after unbalancing key (weighted). # 25.22442 movement pattern cost (weighted). # 0.858196 asymmetric bigram cost (weighted). # 61.54082 irregularity cost (weighted). # 14.56136 manually assigned bigram penalty (weighted) # 11.26684 unbalancing key after neighboring finger (weighted) # 0.858196 asymmetrc bigrams (weighted) # 63.98115 inconsistent symmetries of similar keys (weighted) And Bone # Evolved Layout jduax phlmwqß ctieo bnrsg⇘ fvüäö yz,.k # 13.17988 x100 total penalty per letter # 48.23482 x10 billion total penalty compared to notime-noeffort # 7.653800 mean key position cost in file 1gramme.txt ( 56.02169836 ) # 13.62371 % finger repeats in file 2gramme.txt ( 65.76721351617212 ) # 11.19343 million keystrokes disbalance of the fingers ( 22.38687481869045 ) # 0.382883 % finger repeats top to bottom or vice versa ( 9.241662416541908 ) # 2.569885 % of trigrams have no handswitching (after direction change counted x 1 ) ( 11.367413861129988 ) # 9.573072 billion (rows²/dist)² to cross ( 124.44994845229935 ) # 0.038026 hand disbalance (4% shift is too much). Left: 0.5380261326920235 %, Right: 0.4619738673079765 % ( 1.1133233161947422 ) # 0.731946 badly positioned shortcut keys (weighted). # 26.70450 no handswitching after unbalancing key (weighted). # 29.34707 movement pattern cost (weighted). # 0.868678 asymmetric bigram cost (weighted). # 72.12450 irregularity cost (weighted). # 15.55708 manually assigned bigram penalty (weighted) # 12.39133 unbalancing key after neighboring finger (weighted) # 0.868678 asymmetrc bigrams (weighted) # 34.27499 inconsistent symmetries of similar keys (weighted) And Dvorak # Evolved Layout ’,.py fgcrl/= aoeui dhtns- ;qjkx bmwvz # 15.14281 x100 total penalty per letter # 55.41863 x10 billion total penalty compared to notime-noeffort # 11.55219 mean key position cost in file 1gramme.txt ( 84.55588602 ) # 15.64496 % finger repeats in file 2gramme.txt ( 75.52463069515045 ) # 12.67638 million keystrokes disbalance of the fingers ( 25.352761658066257 ) # 1.892160 % finger repeats top to bottom or vice versa ( 45.67115432320526 ) # 2.084631 % of trigrams have no handswitching (after direction change counted x 1 ) ( 9.220981994058175 ) # 10.33541 billion (rows²/dist)² to cross ( 134.3604324919504 ) # 0.076779 hand disbalance (4% shift is too much). Left: 0.42322086384553576 %, Right: 0.5767791361544642 % ( 2.24792784400046 ) # 1.463892 badly positioned shortcut keys (weighted). # 23.16154 no handswitching after unbalancing key (weighted). # 24.69154 movement pattern cost (weighted). # 0.785204 asymmetric bigram cost (weighted). # 57.79290 irregularity cost (weighted). # 15.63473 manually assigned bigram penalty (weighted) # 20.29272 unbalancing key after neighboring finger (weighted) # 0.785204 asymmetrc bigrams (weighted) # 33.43002 inconsistent symmetries of similar keys (weighted) And Colemak # Evolved Layout qwfpg jluy;[] arstd hneio` zxcvb km,./ # 16.75777 x100 total penalty per letter # 61.32896 x10 billion total penalty compared to notime-noeffort # 12.14284 mean key position cost in file 1gramme.txt ( 88.87909212 ) # 20.96621 % finger repeats in file 2gramme.txt ( 101.21246864529785 ) # 9.545258 million keystrokes disbalance of the fingers ( 19.090517445126185 ) # 0.691273 % finger repeats top to bottom or vice versa ( 16.68530444339314 ) # 5.250280 % of trigrams have no handswitching (after direction change counted x 1 ) ( 23.223643316612847 ) # 14.23883 billion (rows²/dist)² to cross ( 185.10481378085862 ) # 0.048572 hand disbalance (4% shift is too much). Left: 0.45142727602692573 %, Right: 0.5485727239730742 % ( 1.4221048079827046 ) # 0.365973 badly positioned shortcut keys (weighted). # 14.59635 no handswitching after unbalancing key (weighted). # 29.06170 movement pattern cost (weighted). # 0.798238 asymmetric bigram cost (weighted). # 81.11430 irregularity cost (weighted). # 7.396358 manually assigned bigram penalty (weighted) # 16.40680 unbalancing key after neighboring finger (weighted) # 0.798238 asymmetrc bigrams (weighted) # 27.93192 inconsistent symmetries of similar keys (weighted) And Workman # Evolved Layout qdrwq jfupüöß ashtg yneoiä zxhcv kl,.' # 14.36889 x100 total penalty per letter # 52.58630 x10 billion total penalty compared to notime-noeffort # 12.89668 mean key position cost in file 1gramme.txt ( 94.39681132 ) # 17.86578 % finger repeats in file 2gramme.txt ( 86.24539778590645 ) # 12.08234 million keystrokes disbalance of the fingers ( 24.164686943570228 ) # 0.687770 % finger repeats top to bottom or vice versa ( 16.600740467988757 ) # 5.599926 % of trigrams have no handswitching (after direction change counted x 1 ) ( 24.77023814685577 ) # 9.109401 billion (rows²/dist)² to cross ( 118.42221979182285 ) # 0.031606 hand disbalance (4% shift is too much). Left: 0.46839386595704513 %, Right: 0.5316061340429549 % ( 0.925359574421808 ) # 0.731946 badly positioned shortcut keys (weighted). # 9.121154 no handswitching after unbalancing key (weighted). # 29.38692 movement pattern cost (weighted). # 0.861136 asymmetric bigram cost (weighted). # 44.84469 irregularity cost (weighted). # 14.14287 manually assigned bigram penalty (weighted) # 8.771001 unbalancing key after neighboring finger (weighted) # 0.861136 asymmetrc bigrams (weighted) # 52.47782 inconsistent symmetries of similar keys (weighted) And Capewell # Evolved Layout .mydg ;wh,'äü aresf ktnioö xczvj bpluq # 17.17380 x100 total penalty per letter # 62.85149 x10 billion total penalty compared to notime-noeffort # 9.273710 mean key position cost in file 1gramme.txt ( 67.87857252 ) # 15.65823 % finger repeats in file 2gramme.txt ( 75.58866973441775 ) # 9.969009 million keystrokes disbalance of the fingers ( 19.938019290028716 ) # 1.938001 % finger repeats top to bottom or vice versa ( 46.77760462554652 ) # 5.055847 % of trigrams have no handswitching (after direction change counted x 1 ) ( 22.363605543022903 ) # 11.21807 billion (rows²/dist)² to cross ( 145.8349928354487 ) # 0.021514 hand disbalance (4% shift is too much). Left: 0.5215148859918165 %, Right: 0.4784851140081835 % ( 0.6299095523059971 ) # 0.365973 badly positioned shortcut keys (weighted). # 19.27046 no handswitching after unbalancing key (weighted). # 56.78736 movement pattern cost (weighted). # 0.868058 asymmetric bigram cost (weighted). # 59.98423 irregularity cost (weighted). # 15.15501 manually assigned bigram penalty (weighted) # 23.89540 unbalancing key after neighboring finger (weighted) # 0.868058 asymmetrc bigrams (weighted) # 73.17709 inconsistent symmetries of similar keys (weighted) And Carpalx QGMLWY # Evolved Layout qgmlw byv;äöß dstnr iaeohü zxcfj kp,.' # 16.58107 x100 total penalty per letter # 60.68228 x10 billion total penalty compared to notime-noeffort # 12.18032 mean key position cost in file 1gramme.txt ( 89.15340072 ) # 26.57325 % finger repeats in file 2gramme.txt ( 128.27992521054068 ) # 11.39203 million keystrokes disbalance of the fingers ( 22.78407179197142 ) # 0.715443 % finger repeats top to bottom or vice versa ( 17.268676840693004 ) # 1.693749 % of trigrams have no handswitching (after direction change counted x 1 ) ( 7.491984875338312 ) # 10.74563 billion (rows²/dist)² to cross ( 139.69320958006298 ) # 0.018099 hand disbalance (4% shift is too much). Left: 0.5180999176434731 %, Right: 0.481900082356527 % ( 0.5299266295862407 ) # 0.365973 badly positioned shortcut keys (weighted). # 23.29993 no handswitching after unbalancing key (weighted). # 20.57280 movement pattern cost (weighted). # 0.860685 asymmetric bigram cost (weighted). # 64.15116 irregularity cost (weighted). # 13.27022 manually assigned bigram penalty (weighted) # 28.53984 unbalancing key after neighboring finger (weighted) # 0.860685 asymmetrc bigrams (weighted) # 50.56105 inconsistent symmetries of similar keys (weighted) Zwei aktuelle Ergebnisse des Optimierers sehen so aus: # Evolved Layout jlüäö wbdgvzß crieo mntsh⇘ yxuaq pf,.k ┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬──────┐ │ ^ │ 1 │ 2 │ 3 │ 4 │ 5 │ 6 │ 7 │ 8 │ 9 │ 0 │ - │ ` │ ← │ ├───┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬────┤ │ ⇥ │ j │ l │ ü │ ä │ ö │ w │ b │ d │ g │ v │ z │ ß │ Ret│ ├─────┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┐ │ │ ⇩ │ c │ r │ i │ e │ o │ m │ n │ t │ s │ h │ ⇘ │ ´ │ │ ├────┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴───┴───┤ │ ⇧ │ ⇚ │ y │ x │ u │ a │ q │ p │ f │ , │ . │ k │ ⇗ │ ├────┼───┴┬──┴─┬─┴───┴───┴───┴───┴───┴─┬─┴──┬┴───┼────┬────┤ │Strg│ Fe │ Alt│ Leerzeichen │ M4 │ Fe │ Me │Strg│ └────┴────┴────┴───────────────────────┴────┴────┴────┴────┘ # 10.78210871353363 x100 total penalty per letter # 39.459616909194295 x10 billion total penalty compared to notime-noeffort # 7.572413917291403 mean key position cost in file 1gramme.txt ( 55.42599508 ) # 12.519519835088733 % finger repeats in file 2gramme.txt ( 60.436817503347655 ) # 11.27532823083481 million keystrokes disbalance of the fingers ( 22.55065646166962 ) # 0.2898099872779524 % finger repeats top to bottom or vice versa ( 6.995153784842009 ) # 2.8155202166115454 % of trigrams have no handswitching (after direction change counted x 1 ) ( 12.453931578375899 ) # 7.876930641687576 billion (rows²/dist)² to cross ( 102.40009834193847 ) # 0.02096796339028617 hand disbalance (4% shift is too much). Left: 0.5209679633902862 %, Right: 0.47903203660971383 % ( 0.6138968357521171 ) # 0.7319461889614385 badly positioned shortcut keys (weighted). # 14.394287021095723 no handswitching after unbalancing key (weighted). # 30.62285244503868 movement pattern cost (weighted). # 0.8442869700654114 asymmetric bigram cost (weighted). # 53.42856826218681 irregularity cost (weighted). # 14.822616052539642 manually assigned bigram penalty (weighted) # 12.220180694883894 unbalancing key after neighboring finger (weighted) # 0.8442869700654114 asymmetrc bigrams (weighted) # 6.654881871245574 inconsistent symmetries of similar keys (weighted) # Finger load %: 8.8 8.3 8.3 18.5 - 14.0 9.4 8.0 9.0 # Evolved Layout ßuobj vwdlzxq aienc gstrm⇘ äüöpy kf,.h ┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬──────┐ │ ^ │ 1 │ 2 │ 3 │ 4 │ 5 │ 6 │ 7 │ 8 │ 9 │ 0 │ - │ ` │ ← │ ├───┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬────┤ │ ⇥ │ ß │ u │ o │ b │ j │ v │ w │ d │ l │ z │ x │ q │ Ret│ ├─────┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┐ │ │ ⇩ │ a │ i │ e │ n │ c │ g │ s │ t │ r │ m │ ⇘ │ ´ │ │ ├────┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴───┴───┤ │ ⇧ │ ⇚ │ ä │ ü │ ö │ p │ y │ k │ f │ , │ . │ h │ ⇗ │ ├────┼───┴┬──┴─┬─┴───┴───┴───┴───┴───┴─┬─┴──┬┴───┼────┬────┤ │Strg│ Fe │ Alt│ Leerzeichen │ M4 │ Fe │ Me │Strg│ └────┴────┴────┴───────────────────────┴────┴────┴────┴────┘ # 10.888789880674773 x100 total penalty per letter # 39.85004127780889 x10 billion total penalty compared to notime-noeffort # 7.336629319184707 mean key position cost in file 1gramme.txt ( 53.7001787 ) # 9.394714996891675 % finger repeats in file 2gramme.txt ( 45.35211280002587 ) # 11.365924557333908 million keystrokes disbalance of the fingers ( 22.731849114667817 ) # 0.3374395089591601 % finger repeats top to bottom or vice versa ( 8.144789213171713 ) # 3.590146875034434 % of trigrams have no handswitching (after direction change counted x 1 ) ( 15.880348957965833 ) # 8.358853723328869 billion (rows²/dist)² to cross ( 108.6650984032753 ) # 0.028071710516587856 hand disbalance (4% shift is too much). Left: 0.5280717105165879 %, Right: 0.4719282894834122 % ( 0.8218792612098086 ) # 1.463892377922877 badly positioned shortcut keys (weighted). # 12.24942377308282 no handswitching after unbalancing key (weighted). # 37.50418312639172 movement pattern cost (weighted). # 0.8815744638602886 asymmetric bigram cost (weighted). # 48.83104583612258 irregularity cost (weighted). # 16.278126137819385 manually assigned bigram penalty (weighted) # 11.730832183324068 unbalancing key after neighboring finger (weighted) # 0.8815744638602886 asymmetrc bigrams (weighted) # 14.265078429248844 inconsistent symmetries of similar keys (weighted) # Finger load %: 10.5 8.6 13.6 11.8 - 11.6 9.4 9.2 9.7 Wichtig hier: CRY war ein Top-Ergebnis des Optimierers, ist aber durch Integration der Erfahrung der letzten 7 Jahre inzwischen 30% schlechter als das aktuelle Top-Ergebnis. Nur um mal mit der Vorstellung aufzuräumen, wir könnten durch 10% Unterschied im Optimierer wirklich was sagen, wenn ein Ergebnis aus dem aktuellen Optimierer kommt und eins andere Kriterien zur Optimierung genutzt hat, selbst dann, wenn der Großteil der Kriterien ähnlich ist. Am Ende muss ich damit Effchen recht geben: Wir werden den letzten Schliff händisch machen müssen, weil kein Optimierer alles erfassen kann. Er kann uns nur sagen, was wir auf keinen Fall machen sollten (Vertauschungen, durch die die Kosten deutlich steigen), aber nicht, welche Änderungen wir maches sollten. Beim ersten ließe sich zum Beispiel fast schmerzfrei CTRL-v nach links unten und das u nach oben bekommen: $ ./check_neo.py --fingerstats -v --check-string 'jluaq wbdgyzß crieo mntsh⇘ vxüäö pf,.k ' # Evolved Layout jluaq wbdgyzß crieo mntsh⇘ vxüäö pf,.k ┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬──────┐ │ ^ │ 1 │ 2 │ 3 │ 4 │ 5 │ 6 │ 7 │ 8 │ 9 │ 0 │ - │ ` │ ← │ ├───┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬────┤ │ ⇥ │ j │ l │ u │ a │ q │ w │ b │ d │ g │ y │ z │ ß │ Ret│ ├─────┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┬──┴┐ │ │ ⇩ │ c │ r │ i │ e │ o │ m │ n │ t │ s │ h │ ⇘ │ ´ │ │ ├────┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴─┬─┴───┴───┤ │ ⇧ │ ⇚ │ v │ x │ ü │ ä │ ö │ p │ f │ , │ . │ k │ ⇗ │ ├────┼───┴┬──┴─┬─┴───┴───┴───┴───┴───┴─┬─┴──┬┴───┼────┬────┤ │Strg│ Fe │ Alt│ Leerzeichen │ M4 │ Fe │ Me │Strg│ └────┴────┴────┴───────────────────────┴────┴────┴────┴────┘ # 10.72001 x100 total penalty per letter # 39.23238 x10 billion total penalty compared to notime-noeffort # 7.384786 mean key position cost in file 1gramme.txt ( 54.05266154 ) # 12.51223 % finger repeats in file 2gramme.txt ( 60.40163007388642 ) # 11.27228 million keystrokes disbalance of the fingers ( 22.54457922223362 ) # 0.292919 % finger repeats top to bottom or vice versa ( 7.07020833284121 ) # 3.433577 % of trigrams have no handswitching (after direction change counted x 1 ) ( 15.187794686758446 ) # 7.856102 billion (rows²/dist)² to cross ( 102.12933428238193 ) # 0.022789 hand disbalance (4% shift is too much). Left: 0.5227890716545408 %, Right: 0.47721092834545914 % ( 0.6672149659004115 ) # 0.365973 badly positioned shortcut keys (weighted). # 15.43280 no handswitching after unbalancing key (weighted). # 31.06422 movement pattern cost (weighted). # 0.843105 asymmetric bigram cost (weighted). # 50.49074 irregularity cost (weighted). # 14.97846 manually assigned bigram penalty (weighted) # 10.44025 unbalancing key after neighboring finger (weighted) # 0.843105 asymmetrc bigrams (weighted) # 6.654881 inconsistent symmetries of similar keys (weighted) # Finger load %: 8.9 8.3 8.3 18.5 - 14.0 9.4 8.0 8.9 (was hier auch sichtbar wird, ist dass der Optimierer mit den 700 Layouts, die er bisher gefunden hat, noch ein gutes Stück vom globalen Minimum entfernt ist) Liebe Grüße, Arne -- Unpolitisch sein heißt politisch sein ohne es zu merken _______________________________________________ Diskussion mailing list -- diskussion@neo-layout.org To unsubscribe send an email to diskussion-le...@neo-layout.org
