3 ms·
> So Bell System decided to conduct their own research, and after testing many different layouts, decided to choose the one that we all use now. The article gl
by weinzierl 9y ago
> So Bell System decided to conduct their own research, and after testing many different layouts, decided to choose the one that we all use now.
The article glosses over the fact that the layout we use now was not not the best in the Bell study. As you can see from the image in the article [1] the traditional circular layout was the only one that had both significantly shorter keying time and significantly lower error rate. The other source they link to [2] says that "Performance and preference differences, though, were deemed to be fairly small overall across the five finalists, so for engineering reasons Bell went with the layout we know today."
[1] https://archive.org/stream/bstj39-4-995#page/n3/mode/2up https://archive.org/stream/bstj39-4-995#page/n3/mode/2up
[2] https://99percentinvisible.org/article/squaring-circle-seventen-telephone-keypad-layouts/ https://99percentinvisible.org/article/squaring-circle-seven...
- abritinthebay 9y agoThough it should also be stated the the circle was the previous commonly used design (rotary phones) so even that test wasn't in isolation...
- thaumasiotes 9y ago> As you can see from the image in the article the traditional circular layout was the only one that had both significantly shorter keying time and significantly lower error rate. The image in the article shows nothing of the kind, as you might notice if you ask "lower than what?" On your analysis, layout IV-A had a "significantly shorter keying time" than layout VI-A, but that's nonsense -- those are the same layout. The answer is that "significantly lower error rate" and "significantly shorter keying time" refer to statistics taken within a comparison group of three layouts, and don't apply to comparisons between a layout from one group and another layout (or the same layout!) from a different group.
- weinzierl 9y ago> The image in the article shows nothing of the kind, You are right in that the comparison was only between the group of three layouts in a row. From the image no conclusion about the performance of the layouts in different rows can be drawn. However, in the same study, they did a direct comparison between the grid and and the circular layouts and here are the numbers: KEYING TIME PER CENT (SECONDS) ERRORS THREE-BY-THREE PLUS ONE 6.01 2.5 TELEPHONE 5.90 2.0 SPEEDOMETER 5.97 3.0 The study concludes that the differences are not significant. So my point still stands: The grid was not chosen because of superiority in the human factors study but because, in the words of the study: "the rectangular arrangements [..] offered certain engineering advantages".