Recent Content

200 wpm - final reflection on Semimak

posted on 2022-07-17 22:17:55

In less than a year of using Semimak, I hit 200 words per minute on it. On my previous layouts the highest I had gotten was 160 wpm, so this is a big achievement for me. It also means that I'm the first and only person to have achieved 200 wpm on their own layout^1 Given this milestone, I thought I should write a little bit about my experience getting here and my final thoughts on the layout.

Semimak is a great layout... well, for me at least! This definitely makes sense, given it was designed around problems that I identified with my experience in other layouts, and it was closely modeled around my personal finger dexterity. If there is anything we should be able to agree on in keyboard layouts, it's that what is "good" is entirely subjective. People have different hands. They have different ideas of what is comfortable, and different physical limits. My ring and pinky fingers are more capable than many, while my middle fingers can be considered pretty slow. Therefore, it's likely that my layout won't be the best experience for a lot of people, despite being great for me!

This idea probably could have been communicated better given that in August of 2020, we held a month-long tournament with an $850 prize pool to see who could learn the layout the fastest.^2 This likely gave many people the impression that Semimak was the "best layout", despite such a thing, of course, not existing. Be skeptical of anyone who claims their layout to be "objectively" better than other layouts. I will not name names, but if someone claims that their layout has 130% higher "efficiency" than another layout, they probably don't know what they're talking about. And it's not a meaningful metric that you should pay attention to.

To me, Semimak is the best example in proving that layout quality is subjective. I love the layout, and truly think that no other layout currently in existence is better for me. There are several other people who still use and like the layout too - notably, Fenno has over 220 wpm on the layout^3 However, there were many people during the tournament who did not like the layout. It was uncomfortable for them, unintuitive, or physically the layout was just too much from them.

"But the analyzer said it was better than every other layout!" Yeah, of course it did. If you make an analyzer optimize a layout for certain metrics, it's going to do better than other layouts in those metrics. What matters is not how well the layout fits the metrics, it's how well the metrics fit you.

The point of Semimak was to be novel. Based off my own experience, I saw a hole in what kinds of things people and analyzers were looking for in layouts. I wanted to add another way to think about layouts, and I succeeded in doing that by making Semimak. I feel satisfied seeing people talk about dSFB, or same-finger skipgrams so frequently when analyzing layouts these days. My work has definitely had some impact in the community, and I'm happy about that.

Semimak, ultimately, is a proof of concept. It also happens to be a pretty good layout, for the right kind of person.

If you take away anything from this post, it's that Dvorak is the most perfect layout ever created and everybody should switch to it right n

Footnotes

Heres a video of me typing almost 200 wpm on Semimak.

Check out Jashe's video documenting the tournament!

Check out Fenno's 220 wpm handcam!

Reflecting on Semimak, 3 months later

posted on 2021-10-14 22:45:02

On July 1st 2021, 12:00 P.M PDT, Semimak was released. NotGate and I held a tournament with an $850 prize pool, and over 200 people participated in it (including me, just for fun). I got a lot of feedback from this!

Problems

Right Index Finger

Despite the relatively low usage of the right index finger, some people had difficulty with it. It is the finger with the highest unweighted movement, but that's difficult to avoid, considering the index fingers have double the responsibility of a normal finger.

Ring Finger Movements

This seemed to be the biggest issue people had with the layout overall. lr' and oa. columns will cause a lot of movement, and the ring fingers are either the slowest or second slowest fingers for people.

This highlights a very interesting effect of relative dexterity. I personally never had any issue with it, which is unsurprising considering I made the layout adapted to my own finger speed. I believed at the time that people could adapt to comfort with this, like I was able to with my pinkies. However, although almost everyone was able to adapt comfortably to the pinky movement, a not insignificant amount of people struggled with the ring fingers.

I now believe that I am simply lucky to have more dexterous ring fingers than average. This would have been difficult for me to know at the time, since no layouts that I had known of at the time had really pushed the limits on the ring fingers. Regardless, this was my biggest oversight with Semimak.

'r and 'll

This is one of the more specific egregious problems with Semimak. Words like you're, you'll, and I'll are abyssmal on the layout because they include bad SFBs on the left ring finger. Though they can pretty easily be alt-fingered, I still found them quite problematic. Because of this, I created my own mod, Semimak JQ, about 2 months after its original release.

f l h v z  ' w u o y
s r n t k  c d e a i
x j b m q  p g , . /

It only makes two swaps, a 3 key cycle, so it is quite easy to learn from the original layout. After about a month of using it full-time, I can confidently say that it's pretty much objectively better than the reference layout, at least for me.

Successes

By the end of the one month tournament, over 5 people had gotten above 100 wpm with the layout on a very difficult typing test - top 10k English words, random punctuation, and 2 minutes. This is really impressive!

However, the more interesting thing is to watch their handcams (for example, Octahedron's; the fingers look just as low movement as the layout set out to be. This is a great success, because it validates my theory of finger movement. This feels accurate for me as well, even at over 160 wpm on Semimak, my fingers feel like they are moving very slowly, especially when compared to other alt layouts.

Today, there are still quite a few using Semimak, several of which can type over 160 wpm on a basic 60 second test. For the people who Semimak works for, it seems to work really well.

Takeaways

Overall, Semimak was a very novel layout. It was the first of its kind, and because of that was extremely different from most layouts at the time, and was overall a risky move.

Semimak did have its shortcomings, and it isn't for everyone. But I think it's fair enough to say that it did spark a new paradigm for new layouts to be made in the style of. The design behind Semimak has been an inspiration for many new layouts created since.

So, though I don't think that everyone should learn Semimak, it is 100% a great success to me. And I could not have done it without the wonderful, welcoming, and unfathomably nerdy layout community, who encouraged and inspired me immensely. Thank you all!

Semimak

posted on 2021-07-01 22:59:40

The Semimak layout is designed to have low finger movement rate - decreasing the overall speed at which your fingers must travel to type on average. This doesn’t only mean to optimize for reducing same-finger bigrams (e.g ed on QWERTY), but disjointeds as well (e.g the m_y in may on QWERTY). Semimak was the pioneer of DSFB and weighted finger dexterity.

Development

I created my own analyzer, genkey, to put my ideas into effect. At the time, it was only made for myself, and was difficult to use. However, many people have used it since then, creating their own layouts with it, and it has been massively improved since its creation. If you're interested in analyzing layouts, or creating your own, give it a shot!

Finger Speed

Most layout analyzers at the moment report an SFB percentage - how many times you use the same finger to hit two keys in a row. But I found this somewhat superficial - not all SFBs are equally bad. They can be heavily influenced by what finger is being used and how much distance is traveled in the SFB. In addition, the statistic can be generalized more. Why do the two keys in the bigram need to be concurrent to be significant?

Skipgrams

Instead of only punishing same-finger bigrams weighted by finger and distance, I had the idea to also punish same-finger skipgrams. I weighted this exponentially - separated by 0 keys (normal bigram) is a weight of 1, separated by 1 is half as bad, separated by 2 is a quarter, and so on. These weights are arbitrary, but they were inspired both by the experiences of both myself and typists much faster than me.

Distance weighting

I weighted the distance between keys exponentially, as I think a distance of 2 units is much worse than a distance of 1 unit. Layouts like Workman and Colemak DH have shown the discomfort that can be cause with lateral index movement - because of this, lateral movement was treated worse in my analyzer.

Finger weighting

I used a basic script that NotGate wrote to test how quickly I can move each of my fingers. For example, to see how dexterous my middle finger is, I would type eded over and over on QWERTY, and see how fast my keys per second was. This data was then used to weight how bad fast distance is on each finger.

Scoring

The layout was scored by genkey largely with the weighted finger movement speed metric. There was also some reward for rolling, and penalty for redirects.

However, what the generator created had an extremely high redirect rate, and so I swapped the columns around at the end to increase alternation. Though I am no alternation lover, the redirect rate would have really been way too much.

Download

Downloads for Windows native and EPKL, MacOS and Linux are available in the Github Release page!

This blog covers typing, layouts

View content from 2021-07, 2021-10, 2022-07


Unless otherwise credited all material Creative Commons License by semi