eggie

After switching back and forth between Dvorak and QWERTY for multi-hour work sessions maybe three times I found that I could do so instantly. I learned touch typing on QWERTY very young so maybe this is why it was easy for me. I often use a keyboard without any labels on the keys so it's always touch typing no matter the layout I pick. Also, I don't really get anxious about things that slow down my text entry, so I tolerated my own interface experiments without worry. I've coded and done data analysis using my thumbs on a cell phone. Switching keyboard layouts really isn't so bad by comparison. To each their own.

I find it surprising that much of the current effort in keyboard layouts is in finding things that are better than QWERTY but similar enough to it to not be scary. I am completely unmotivated by this. If someone could convince me that their layout was a kind of super-Dvorak, and scientifically enhanced the same features that I enjoy about it then I would try it out in a heartbeat.

Did you consider using an isomorphic layout like the typematrix? I find the anachronistic skew in keyboards to be a source of discomfort.

The author doesn't even learn dvorak before deciding that it doesn't work for them. They mention don't like how it behaves on the consonant-heavy, qwerty-optimized commands used in unix. (I have to say this is a fair evaluation, but in practice the layout somehow doesn't encourage RSI despite the apparent awkwardness of typing unix commands on it.)

In result they don't seem to understand that the dvorak layout is optimized for alternation between hands (they do mention this) and inward rolling motion (they don't seem to understand this). You can really feel this when typing, and it makes typing downright enjoyable.

I wouldn't be surprised if much of the added distance relative to colemak and workman is probably reaching for the "i" with the left pointer finger. These are the easiest movements to make repeatedly, and in my experience the home row horizontal movement that they try to optimize out of workman isn't really important.

People should be free and encouraged to make their own layouts to beat RSI. But, scientific measurement of what is better seems basically impossible...

The genome is akin to the software of an organism. It's not exactly obfuscated, so we can actually learn a lot just by reading it. The next big benefits are all about relating the genome of one organism to others.

A single "reference" genome can serve as a high quality basis for interpreting new genomes. This is called resequencing. Small pieces of the new genomes are sequenced with very cheap techniques and mapped into the most-similar bits of the big genome, in effect allowing for a kind of guided reconstruction of the new genome. The short read lengths make it difficult to reconstruct a new genome de novo, but that's OK because if we have a reference we've already paid that price once for a related individual.

This process has bias but it's also expedient, say when you have thousands or millions of genomes in a study (as happens in agriculture).

One big problem in the field now is this obsession with reference genomes and resequencing against them. Researchers forget how much the reference genome can bias their resolution of a new genome. This has resulted in a lot of fussing and fixation on high-entropy regions of genomes and small variations (point mutations, or SNPs) within them. There is a growing body of evidence that suggests that large changes are probably more important for adaptation, but they remain somewhat underappreciated because of this short read resequencing modality.

No. You need a wheat cell as well. Then you might be in business.

And, no, it is not possible (today) to rebuild a multi gigabase hexaploid genome. The best that's been done so far is mycoplasma (500kb genome, which is many orders of magnitude smaller).

Also, the "genome sequence" presented here is not really complete. Thus far only the simplest organisms have had their genomes synthesized from a digital template and resulted in a viable cell. We rarely can produce a "complete" genome for anything other than bacteria and viruses. And for those, the diversity in the population makes the idea of establishing a single genome sequence somewhat limiting.

I look for all of these things as well, and they are critical. But none of them have any relevance in a technical domain if the person who has them isn't capable of pulling more weight than anyone else. Management is only approximately a learned and transferable skill. At the end of the day, lack of capacity in the target domain will make any managerial knowhow useless.

Also to be clear, I'm not talking about software development in the mythical man month sense, but systems thinking and a facility with abstractions and math combined with the ability to make such ideas real. Such people have the power to lift everyone up to their level, guide the group along an easy path, and fill in the gaps in ability present in their team.

There is a reason that people learn martial arts from people who have developed skill at them, not those who are good at managing workouts. The latter will only teach you how to manage your workouts. If you want to learn an art, learn from a master. In technical domains the masters are the best leaders, and you learn from them in every sense.

In my experience it is rarely enough to just "lead" more thoughtfully, which seems to be what the author here is promoting. Software development does not just require servant leaders in the helping sense, but in the working sense. The best leaders in software and technical domains are also those who can write (either code or ideas) better than anyone else. They have the power to move the project forward on their own, and working with them lightens the effort for everyone. Those who also adopt the ideas in this essay do foster the sensation among those they lead that they were never led at all, and rather achieved their common objectives together.

Using this on an OLED screen really makes it clear that the current standard of light backgrounds and dark text is only optimal with backlit LCD panels.

Any 3-color screen will never be true to life accurate. It's always a simulation of the real thing projected into the three dimensions of color that most of us perceive.

Colors appear to change dramatically under different lighting conditions, but we accept these changes as normal given context. Changing the white balance on your screen is no different. It's not inaccurate, it's just rendered as if in different lighting conditions.

It's no worse than on a phone. I've never noticed burn in on any OLED phone or my current laptop (although the latter is newer).

I'm using a thinkpad X1 yoga OLED. The biggest danger with this device is using Windows, as the drivers for the display panel have a bug in them that will lead to degradation and destruction of the panel. Using Linux avoids this issue.

For personal red illumination I use USB-rechargeable NightRider rear bike lights. I ultimately got two and I take them everywhere with me. They are great night lights, especially when travelling, except for the fact they flash blue when you turn them off to indicate their charge level! Using bike lights makes sense as they are designed to be durable, waterproof, and long lived.

I've used a series of OLED screened phones. Rooting can help to ensure you really can put a full screen temperature adjustment on the phone. I regret not having done this on my current phone, which is a oneplus model.

In the past year or two a huge number of multicolor LED lighting solutions [1] have shown up on amazon. These are usually cheap strips of several hundred multicolor LEDs, and come with an IR remote that lets you set one of a number of colors--- among which are pure rend and oranges that are nice in the evening.

[1] For instance: https://www.amazon.com/TOPMAX-16-4ft-Strips-Lighting-Charger...

Ubuntu 18.04 has a night mode built into the gnome shell. It works really well, allowing a complete removal of blue.

But neither of these has any point if you have blue light bleeding through from the LED backlight of the LCD screen. The most complete solution is OLED and a screen temperature adjustment like this.

After realizing the effect that blue light had on my sleep I went on a crusade to eliminate it from my life after sundown.

I used red LED bike lights for room illumination until I found a series of alternatives. The cheap, remote-controllable 5 meter multicolor LED strips are an amazing solution to evening lighting. It is also possible to get "warmer" LED bulbs. These are often marketed as looking like antiquated filament bulbs. But the best solution is pure red light.

I also removed every single blue-LED backlit computing device from my life. I've made sure to use OLED phones since they were first available. I finally have an OLED laptop, and thankfully, as they are no longer being made! The change in laptop screen really is the most transformative thing for me, as I often need to work in the evenings. I set all the backgrounds in the OS to black, and use browser plugins to set the background of the web to black as well. I have never slept better since blue LEDs came into my life.

People really need to take this issue seriously. There is evidence that blue light's disruption of circadian rhythms and suppression of melatonin production is related to cancer and other health risks. Blue should be reserved for the color of the sky, which is where our bodies evolved to recognize it directly as the input to our circadian rhythms.

I am pleased by the wave of "night light" filters for phones. This is a sign that people are catching on. They use the filters and feel able to get to sleep. We need to go further. OLED everywhere and deep night mode for all.

I've watched many of my and coworkers spending huge amounts of time (many hours a month) fiddling with the complexities of managing software packages on OSX systems. Everything was so frighteningly manual and impossible to reproduce. They easily individually used 50x as much time doing that as I spent fiddling with my Linux installation. Those are levels of effort that definitely matter to me enough that I put importance in the OS I use.

It's nice when things fit in RAM. But very often they don't. When you want to sort arbitrary-sized binary records on disk look no further than bsort: https://github.com/pelotoncycle/bsort

I had the same kind of problem. For me resolving the issue was as simple as learning and using the dvorak layout. Having a different and entirely more efficient way to type completely resolved the problems I was having. This in combination with a typematrix keyboard has ensured that I never once experienced hand pain in more than ten years working with code.

Before, when using QWERTY on a regular keyboard, I reached a point where I could no longer type at all without pain. Although I code I mostly write prose and messages. After switching I can type on QWERTY without pain as the original injuries healed, but I can feel after several hours how difficult and uncomfortable it is to type on that layout.

If you're at a point where you can no longer type why wouldn't you consider changing the layout of your keyboard? Even if it's to something substandard in many ways it at least ensures you use your hands and arms in different patterns. It's also a nice exercise for the mind.

Your exposition of Dawkins' ideas is neat, and probably helpful to people in an introductory biology course. I'd like to help clarify some of your comments on the genome.

> Even your DNA is not really a unified whole, it's a bunch of genes that have glommed together in a cooperative effort to reproduce themselves, but it's not an organized or homogeneous effort. That's why you have freeloaders ("junk DNA") and defectors (cancer genes) who put their own interests above those of the group.

"Cancer genes" don't exist in the sense that there aren't genes that specifically cause cancer. If there were then every individual with a full gene complement would get cancer. What you're referring to are certain alleles that can arise which change cellular processes so as to break cooperative patterning of cell behavior. If you're lecturing in evolution it might be nice to use a technically correct description to match the Dawkinsian semantic splicing you're resting on :) Call them "oncogenic alleles" instead.

"Junk DNA" is a blind concept that conflates sequence complexity with function. The argument goes that these pieces of DNA are just copies of each other and serve no obvious purpose, thus they are junk that "freeloads" off their host. This ignores the fact that these elements exist in virtually all complex organisms, and have lineages that last tens or hundreds of millions of years, which is longer than the lifespan of most species. If they were only an energetic drag on their host they would be selected out over these time scales. Instead, they appear to serve important purposes in genomes. There is a huge amount of evidence that this "junk" is also an "organ" of the genome that provides essential function in an evolutionary sense. This perspective is old, more than 30 years, and the first reference to it I'm aware of is from a study in 1983 that showed that Escherichia coli with transposons evolved faster than those without: https://www.nature.com/articles/303633a0

What reference model do you align the RNAseq against? That would seem to have a huge effect on results no?

Would this allow rapid realignment of the data against a given reference (graph?) model? Or is the alignment somehow baked in?

> certain DNA bases can shape-shift for a thousandth of a second, transiently morphing into alternative states that can allow the replication machinery to incorporate the wrong base pairs into its double helix

I find this fascinating. I had always assumed that it was the shapeshifting of the polymerase that caused replication errors. Clearly that's not the whole story!

Software is only 1% of the story. We need reliable public data from many people linking genomes, diet, life history, and phenotype to do the other 49%.

That seems expensive. I got a personal sequence done (at 15x) four years ago for 750 usd. It was part of a group deal so maybe it isn't relevant to your search.

The technology really works and is generating results at a rapid pace. This will absolutely not be the next Theranos, although you may be right that there are issues of burnrate and maybe even mismanagement.

To me the fact that they have a patent monopoly on "putting DNA through a protein pore with voltage sensing" is tragic. Who knows were we would be today if these patents had been granted to the public domain.

The per base error rate is bad. In the case of pacbio, this error process approximates white noise, and so you can deal with it perfectly by increasing read coverage. Things are somewhat complicated with the nanopore tech described in this post, as errors may be correlated due to the way the basecalling is done, but in practice it's nearly as big a problem as you think it is.

For things approaching a read length the per-base error rate of a single read is simply irrelevant. In practice, with sufficient coverage (e.g. 20x) you simply don't care about the per base error rate of the reads.

I've participated in several large projects that worked to detect SVs from short reads in humans. The results, which remain best of class, are simply disappointing. A tiny fraction of the variants detected were actually resolvable to near-base pair resolution. The vast majority were described in approximate terms, using estimates of breakpoints and allelic structure.

Most structural variation I've seen based on whole genome assemblies is not even classifiable into neat categories like "deletion" or "insertion". If you think that "most" things are detected with short reads then you are deluded by the dominant technology.

> you can get a full human genome sequenced at high coverage for between 1000 and 3000 USD

This is not a "full" human genome, but a collection of 150bp fragments that can be realigned to an existing human genome. You cannot take this and infer the whole diploid genome of the individual. There is a huge amount that will be missed, and all of our current knowledge is based on this gappy picture of what's going on in single genomes and human populations.

> It can give very long reads, which are useful in some niche applications. But it’s been massively over-hyped (and over capitalized).

I think you're dismissing the technology out of hand because of biases derived from much more limited short-read technology that only allows us to reliably see small variants <50bp.

Without these long reads we can't see structural variation (SVs). There is an increasing amount of evidence that much of adaptive variation is driven by these kinds of variants. If you want recent evidence, see https://www.nature.com/articles/s41588-017-0010-y. There has long been evidence that there are huge copy number variations in humans, but these are still not evaluated reliably: http://science.sciencemag.org/content/330/6004/641.

We should be open to the possibility that our observational techniques are limiting our understanding how how genomes work. This has consistently occurred in the history of every observationally-driven science.

It's amusing to me that people assume that SVs are "niche" when even the limited surveys of genomes we've been able to do with short reads show that roughly an equal number of base pairs in the human population vary due to small variants like SNPs and indels and big ones like deletions, insertions, and large scale copy number variation: http://science.sciencemag.org/content/330/6004/641

This is the elephant in the room. It definitely seems that something is up when $100M in tokens flow into exchanges and the price immediately jumps.

I am trying to find moderate-resolution market data to evaluate the effect that Tether issuances have had on prices. Does anyone know of a public data source that's higher resolution than 24h?

When I was young and impressionable I repeated some of his crazier assertions in public. I regret it to this day.

I'm impressed at his staying power in the public discourse about tech despite how incredibly wrong he has been.

Even if we claim that "real AI" is a system that can actively perform at human levels in creative tasks, I wouldn't believe that it will always be in the future. However, I don't think this is very near, and I think this is sort of what people are implying when they talk about AI. To me saying "general AI is what we're exploring and making mini-discoveries in" is a kind of apology. Unfortunately, this apology pushes us into a grey area where "AI" is really just the scientific artifice. That's not the same as AI that has auto-creative capacities.

So we are breaking off small pieces of a bigger thing. If that thing is AI, does it already have a form and an older name in the world?

Would agree that in this article AI is a stand-in for what used to be called Math, Statistics, and Computer Science?

Our scientific output is an artifice, and it is "intelligent" in its ability to guide our action. To me it isn't then strange to think of the sciences as the "general AI".