The color temperature of a lightbulb is a very poor indicator of how much blue light it gives off as the spectrum of LED bulbs tends to have a spike specifically around blue rather than being a continuous curve as shown in the animation. Two bulbs that have similar or identical color temps can actually have very different spectrums.
There's a subset of "pro-gaming" glasses (famously Gunnar, but there's others) that have been around for a while that purport to do the same kind of blue light blocking, though they typically advertise it as preventing eye strain.
I was gonna say since they're for cool gamer dudes they're not supposed to look nearly as dorky but I hadn't looked at Gunnar's website in a while and dang they have a ton of options for frames now (and even way back when you could get them in prescription).
As someone who's done my fair share of reading about color science/perception, I wouldn't necessarily trust that animation at the end to provide suitable light tempuratures for blue light reduction. It looks like the spectrum calculations are correct, but there are lots of nonlinearities between a 2000K label on a lightbulb and melatonin production in your brain (ex. cone response curves, chromatic adaptation, logarithmic sensory response, light metamerism, etc).
I think you'd probably have to do another optimizing-tea style post but with increasingly red room lights. Good luck doing a double blind on that one haha.
So, in 2017 everyone went crazy about blue light blocking glasses, then in mid 2018 Scott Alexander published Melatonin: Much more than you wanted to know, then by around 2019, blue-blocking glasses seemed to disappear.
Have you seen any reviews on melatonin alone vs melatonin + blue light blocking glasses vs blue light blocking glasses?
I've got a pair of the blue-blocking glasses. I bragged to a friend, "they only cost $5!" His response, "Could you have paid more for less dorky ones?"
My main use for them is adjusting to jet lag, but the timing is critical. I wrote it up, and made a calculator to figure out when to wear them: https://unaging.substack.com/p/jet-lag-hacked
Strangely, it isn't about rods or cones, it's about something called ipGRCs, which are an additional, much less dense layer of sensing in the retina.
A more useful resource is the IES's publication "Simplifying Melanopsin Metrology," which is much more comprehensive than any other source of its length.
For a simple intervention, I recommend the Hue bulbs, with "adaptive lighting" in homekit. While 5000K → 2700K is only a 50% drop, you also want to decrease brightness automatically—likely from a full 800-1600 lumens in the day, down to around 100 at night. Together, you're looking at something like 3% of the melanopsin output you'd otherwise have, even ignoring the effects of sunlight.
Amazing, thank you! How skeptical should that first curve make me about this whole idea? It looks like the melanopsin-sensitive cells have a really wide spectrum that doesn't really collapse until you get down to pure-red light.
Naively, that tends to make me think that just *dimming* the lights as much as possible is probably the way to go, and color-shifting is pretty minor? Which is... ummm... exactly what you're suggesting. :)
It's the other way around—IPGRCs respond pretty narrowly around 480nm (half-max bandwidth ~40nm), but the lights themselves are designed to have a very even, wide-spectrum.
I wouldn't ignore the color shifting entirely—6500K / 2700K is about a factor of three, and keep in mind that screens have a white point of 6500K. At the margin, it's easier to shift color on a screen at night than to make it 2/3 darker.
The other half of this is that you can also max out daytime melanopsin—daytime lighting probably should be all 6500K, although you need high CRI lights to make this pleasant, ideally violet phosphor pump.
But yes—as suspected, turning lights down or off does indeed make people sleepier.
If I understand it correctly, if I want to limit the melatonin response to 10% of max, I need to make sure I only get frequencies above ~555nm, no? Wouldn't that mean I'm only left with blue/green?
In any case, you've definitely convinced me that your strategy is the wise one. Instead of going insane with color temperature, do some kind of "defense in depth" by combining color and dimness.
"Lockley et al. in 2003[19] showed that 460 nm (blue) wavelengths of light suppress melatonin twice as much as 555 nm (green) light, the peak sensitivity of the photopic visual system", almost certainly due to the retinal melanopsin cells. The wikipedia entry on them is worth reading https://en.wikipedia.org/wiki/Intrinsically_photosensitive_retinal_ganglion_cell
The chart is a little criminal—the blue colored curve on it corresponds to red light. I assume default excel chart settings. You would be left with only yellows and reds.
Very interesting! I have noticed the season has a huge effect on how much artificial lighting affects me. In the summer I can sleep with some ambient light, and I am not bothered by ugly LED lighting - I think the large amount of natural sunlight in the summer drowns out the effects of artificial light. But in the winter, LED lights give me headaches- I always switch them out for incandescent bulbs sometime in late fall. My sleep is much more finicky in winter, and needs total darkness.
This makes me wonder whether the season each experiment was done in, has any effect on the results.
• Shelly Duo RGBW smart bulbs: 2700K-6500K (admittedly pretty far from the full gamut but subjectively they go from "really warm" to "really cold") and 0-200 lux (for a single bulb in a small-to-medium room; I have enough bulbs in my claustrophobic, small-roomed, gaol-cell-like little hovel to provide 0-1000 lux in the bedroom and living room...)
• Controlled by Home Assistant running on a spare old computer (intimidating-looking but ultimately easier to get your head around than you think; even in the worst-case scenario it takes about a weekend to set-up from scratch with no prior knowledge)
• Configured such that a single press of a light switch gives you a daytime(ish) brightness and colour temperature and a double-press gives you sunset(ish) brightness and colour temperature (Could have just as easily had the system choose the colour temperature based on the time of day, or any other criteria, but preferred manual control)
If enough other willing participants had smart setups compatible with some reasonable trial protocol, I think we could fairly easily run a trial here!
(Possible confounding factor: people who run Home Assistant might tend to look at glowing rectangles rather more than the general population..?)
Even for a journal-standard RCT, though, I suspect that compared to costs such as "accommodation for all participants in a sleep lab", or "writing a legally-watertight safety case for depriving participants of sleep for upwards of 40 hours", it might not actually be prohibitively expensive to pay for a handful of smart bulbs and (eg.) raspberry Pis running Home Assistant...
> Even for a journal-standard RCT, though, I suspect that compared to costs such as "accommodation for all participants in a sleep lab", or "writing a legally-watertight safety case for depriving participants of sleep for upwards of 40 hours", it might not actually be prohibitively expensive to pay for a handful of smart bulbs and (eg.) raspberry Pis running Home Assistant...
I wonder about this! Maybe it's because I personally am *terrified* of trying to get something like Home Assistant running. But I'd think you'd need some amount of tech support.
For a study presumably you'd provision dozens of Raspberry Pis and lightbulbs, all running the same software, such that all the participant would have to do would be screw-in (twist-in... nuts to Edison...) the lightbulbs and plug the rPi into a port on the back of their router and it would all Just Work.
The lightbulbs would need to be paired to the participants' WiFi, which couldn't be done in advance by the experimenter, but this is a super basic procedure: for Shelly bulbs, the bulb creates a WiFi hotspot when screwed-in to a light socket, you connect your phone/computer to that hotspot and navigate to a given web address where you can put in the password for your home WiFi. Then, you disconnect from the hotspot and the bulb will connect itself to the home WiFi.
Whereupon the Home Assistant rPi, plugged-in to the router, will automatically see it and include it in the program it's running (which the experimenter can set-up identically on all the participants' rPis beforehand). The bulbs only expose themselves to the local WiFi subnet, not to the internet (for absolutely sound security reasons) but if necessary (and if ethically approved) the Home Assistant rPis, which are basically tiny cute little adorable computers, could be exposed to the internet and could report on each bulb's status in real-time to the experimenter (say to monitor protocol compliance, do statistical wizardry with the data or even just to identify and offer tech. support to -or automatically exclude- any participants whose setups somehow weren't working as intended.)
I'm sure you're already in possession of plenty of unsolicited Home Assistant technical advice so I shan't add my voice to the cacophony - but if you do ever want some stolidly-pedagogical, so-simple-it's-almost-patronising guidance or help in setting it up I'd be happy to: feel free to DM me if so.
For the bulb thing, I have some Hue lights which are nice in the bedroom, but they're really expensive and kind of annoying (you need an app to change them). For my bathroom and a few other part of the house, I got lights that change color temperature when you dim them. It's a little finnicky which dimmers they work with and how many you can add to a circuit, but once they're setup it's nice to just slide the dimmer switch at night. The ones I have are pretty expensive but I imagine someone less perfectionist makes a cheaper version with similar features: https://store.yujiintl.com/products/well24-flamewarm-a19-a60-11w-dimmable-led-sleep-bulb-1250k-2pcs-4pcs?variant=45378286584065
Well, I have very mixed feeling about that! On the one hand, I'm incredibly averse to having anything in my home run on apps, and that's a wonderful "midwit-home" solution, exactly what I'd like.
On the other hand, holy shit, $76 for two bulbs?
I actually have some (cheap) bulbs that cycle color temperature when you flick them on and off. I think someone recommended them after my midwit-home post. But I find them so fiddly that I rarely mess with them.
Just switch bedroom bulbs to smart ones, and set them on a timer to get warmer at night. They are really not that expensive compared to blue-light glasses. And for screens - f.lux will change your life
Yes, I like f.lux also-have used it for a few years on my work computer. Seems to have more capacity than I understand the point of. I keep it set on macular pigment which feels very comforting to my eyes.
One thing that would be easy to do is switch out all bedroom lights to 2700K. I'm thinking of putting a second floor lamp in the room I spend all my time it -- right now I've got a floor lamp with a 5000K LED in there -- but I've got a spare floor floor, so I could just put in a 2700K light in it and switch lamps when/if I think of it. Not ideal, but, still, something.
What's frustrating to me is I don't think 2700K will do *that* much—it seems like it only reduces the fraction of blue light by ~33% compared to 5000K. I want 2000K bulbs!
I went and looked on Amazon, and 2000K LED bulbs are really a thing. They are only equivalent to 60W bulbs (600 Lumens) so pretty dim, but you could plug in 2 or 3. Amazingly enough, they aren't even that expensive -- about $3 a bulb.
Great post, and I'd love to see a follow up. I recently started using blue light blocking glasses during the day while working on my computer to reduce eye-strain migraines, per the recommendation of an eye doctor. I was extremely skeptical because the sun produces way more blue light than any computer screen. But after trying them, eye strain, which occurred every day very bad, is virtually gone. I understand the detailed physics behind light spectra and am very unsusceptible to the placebo effect. But man, they seem to work. I'd love to see a follow-up to this article where you aggregate these studies about blue light blocking glasses and eye strain. Thank you!
I did stumble past quite a number of papers that looked at stuff like that! I was laser-focused on sleep, but I think there's a lot of research out there...
The part about color perception at the beginning reminded me of this recent paper in which the authors get their subjects to experience a new hyper-saturated blue-green color by firing a laser directly into their M cones: https://www.science.org/doi/10.1126/sciadv.adu1052#
FWIW, I have an orange lamp in my room that I turn on at night (normal bulb, transparent orange paper rolled into a cylinder for the shade). I don't use it for any scientific reasons, I just find it soothing.
Oh wow, that is extremely interesting! The idea seems to be that out three different cones all have somewhat overlapping sensitivity to different wavelengths. So we never (normally) experience just one single type of cone firing. But with this you can. Amazing.
Personally I have 2 sets of lights for day and night. 2500 lux of cool light during the day, then a separate set of Philips hue lights for after work + night.
The hues start as 2700k after work, then switch to red around 9pm.
I'm always sad when I go to a hotel and it doesn't have red lights :(
In general my midwit-home principles (https://dynomight.net/midwit-home/) make me super hesitant to try set something like this up. This seems like the strongest reason to consider doing smart-home stuff.
Though AFAIK none of the smart bulbs do color temperatures much below 2700K? I want 1800K!
My lifx bulbs go that well and automatically start going lower in colour temp in the evening. This article has inspired me to be more aggressive with that
I had the same scepticism. I started with just controlling everything via hue. Smart lights plus hue smart plugs to control the strings of bright lights. And hue wall switches. It worked well and was zero maintenance.
Eventually I switched to home assistant which is very fun and I have lots of useful automations. But it does completely break once or twice a year. Only worth it if you like tinkering!
The hue-only setup I do recommend though. Zero maintenance it just works
The color temperature of a lightbulb is a very poor indicator of how much blue light it gives off as the spectrum of LED bulbs tends to have a spike specifically around blue rather than being a continuous curve as shown in the animation. Two bulbs that have similar or identical color temps can actually have very different spectrums.
Yup: https://dynomight.net/blue-light/#:~:text=You%20can%E2%80%99t%20sense
The good news is that the cells that produce melatonin seem to have a similar breadth, so it seems likely that these differences don't matter?
There's a subset of "pro-gaming" glasses (famously Gunnar, but there's others) that have been around for a while that purport to do the same kind of blue light blocking, though they typically advertise it as preventing eye strain.
I was gonna say since they're for cool gamer dudes they're not supposed to look nearly as dorky but I hadn't looked at Gunnar's website in a while and dang they have a ton of options for frames now (and even way back when you could get them in prescription).
As someone who's done my fair share of reading about color science/perception, I wouldn't necessarily trust that animation at the end to provide suitable light tempuratures for blue light reduction. It looks like the spectrum calculations are correct, but there are lots of nonlinearities between a 2000K label on a lightbulb and melatonin production in your brain (ex. cone response curves, chromatic adaptation, logarithmic sensory response, light metamerism, etc).
I think you'd probably have to do another optimizing-tea style post but with increasingly red room lights. Good luck doing a double blind on that one haha.
So, in 2017 everyone went crazy about blue light blocking glasses, then in mid 2018 Scott Alexander published Melatonin: Much more than you wanted to know, then by around 2019, blue-blocking glasses seemed to disappear.
Have you seen any reviews on melatonin alone vs melatonin + blue light blocking glasses vs blue light blocking glasses?
Bugreports/minor errors!
The mistake box on your site (for this post only?) is unavailable, server timeout. Just FYI.
The mistake box renders the text I type as black while the theme is already dark, making it unreadable. (I'm on Chrome on mobile if that matters.)
> Appleman et al. (2013) either exposed people to different amounts
That sentence is missing an "or" or the "either" is wrong
> body might regulate its sleep schedule based that input.
Should be "based on"
I've got a pair of the blue-blocking glasses. I bragged to a friend, "they only cost $5!" His response, "Could you have paid more for less dorky ones?"
My main use for them is adjusting to jet lag, but the timing is critical. I wrote it up, and made a calculator to figure out when to wear them: https://unaging.substack.com/p/jet-lag-hacked
Strangely, it isn't about rods or cones, it's about something called ipGRCs, which are an additional, much less dense layer of sensing in the retina.
A more useful resource is the IES's publication "Simplifying Melanopsin Metrology," which is much more comprehensive than any other source of its length.
https://www.ies.org/fires/simplifying-melanopsin-metrology/
---
For a simple intervention, I recommend the Hue bulbs, with "adaptive lighting" in homekit. While 5000K → 2700K is only a 50% drop, you also want to decrease brightness automatically—likely from a full 800-1600 lumens in the day, down to around 100 at night. Together, you're looking at something like 3% of the melanopsin output you'd otherwise have, even ignoring the effects of sunlight.
Amazing, thank you! How skeptical should that first curve make me about this whole idea? It looks like the melanopsin-sensitive cells have a really wide spectrum that doesn't really collapse until you get down to pure-red light.
Naively, that tends to make me think that just *dimming* the lights as much as possible is probably the way to go, and color-shifting is pretty minor? Which is... ummm... exactly what you're suggesting. :)
It's the other way around—IPGRCs respond pretty narrowly around 480nm (half-max bandwidth ~40nm), but the lights themselves are designed to have a very even, wide-spectrum.
I wouldn't ignore the color shifting entirely—6500K / 2700K is about a factor of three, and keep in mind that screens have a white point of 6500K. At the margin, it's easier to shift color on a screen at night than to make it 2/3 darker.
The other half of this is that you can also max out daytime melanopsin—daytime lighting probably should be all 6500K, although you need high CRI lights to make this pleasant, ideally violet phosphor pump.
But yes—as suspected, turning lights down or off does indeed make people sleepier.
I'm just looking at this figure: https://www.ies.org/wp-content/uploads/2019/07/Simplifying-Melanopsin-Metrology-Fig1.jpg
If I understand it correctly, if I want to limit the melatonin response to 10% of max, I need to make sure I only get frequencies above ~555nm, no? Wouldn't that mean I'm only left with blue/green?
In any case, you've definitely convinced me that your strategy is the wise one. Instead of going insane with color temperature, do some kind of "defense in depth" by combining color and dimness.
"Lockley et al. in 2003[19] showed that 460 nm (blue) wavelengths of light suppress melatonin twice as much as 555 nm (green) light, the peak sensitivity of the photopic visual system", almost certainly due to the retinal melanopsin cells. The wikipedia entry on them is worth reading https://en.wikipedia.org/wiki/Intrinsically_photosensitive_retinal_ganglion_cell
That's all totally consistent with the above figure, though...
The chart is a little criminal—the blue colored curve on it corresponds to red light. I assume default excel chart settings. You would be left with only yellows and reds.
Very interesting! I have noticed the season has a huge effect on how much artificial lighting affects me. In the summer I can sleep with some ambient light, and I am not bothered by ugly LED lighting - I think the large amount of natural sunlight in the summer drowns out the effects of artificial light. But in the winter, LED lights give me headaches- I always switch them out for incandescent bulbs sometime in late fall. My sleep is much more finicky in winter, and needs total darkness.
This makes me wonder whether the season each experiment was done in, has any effect on the results.
After reading the title, I was looking forward to seeing how you managed to perform a double blind experiment on yourself. Thanks for the article!
I have:
• Shelly Duo RGBW smart bulbs: 2700K-6500K (admittedly pretty far from the full gamut but subjectively they go from "really warm" to "really cold") and 0-200 lux (for a single bulb in a small-to-medium room; I have enough bulbs in my claustrophobic, small-roomed, gaol-cell-like little hovel to provide 0-1000 lux in the bedroom and living room...)
• Controlled by Home Assistant running on a spare old computer (intimidating-looking but ultimately easier to get your head around than you think; even in the worst-case scenario it takes about a weekend to set-up from scratch with no prior knowledge)
• Configured such that a single press of a light switch gives you a daytime(ish) brightness and colour temperature and a double-press gives you sunset(ish) brightness and colour temperature (Could have just as easily had the system choose the colour temperature based on the time of day, or any other criteria, but preferred manual control)
If enough other willing participants had smart setups compatible with some reasonable trial protocol, I think we could fairly easily run a trial here!
(Possible confounding factor: people who run Home Assistant might tend to look at glowing rectangles rather more than the general population..?)
Even for a journal-standard RCT, though, I suspect that compared to costs such as "accommodation for all participants in a sleep lab", or "writing a legally-watertight safety case for depriving participants of sleep for upwards of 40 hours", it might not actually be prohibitively expensive to pay for a handful of smart bulbs and (eg.) raspberry Pis running Home Assistant...
> Even for a journal-standard RCT, though, I suspect that compared to costs such as "accommodation for all participants in a sleep lab", or "writing a legally-watertight safety case for depriving participants of sleep for upwards of 40 hours", it might not actually be prohibitively expensive to pay for a handful of smart bulbs and (eg.) raspberry Pis running Home Assistant...
I wonder about this! Maybe it's because I personally am *terrified* of trying to get something like Home Assistant running. But I'd think you'd need some amount of tech support.
That said, I always remember this insane experiment where the trial actually replaced the stoves in a bunch of people's houses: https://dynomight.net/stoves/#:~:text=Except%20heroes%20are%20real%3A
And you may well be right that the other costs of a trial are so high that this wouldn't actually be all that prohibitive...
For a study presumably you'd provision dozens of Raspberry Pis and lightbulbs, all running the same software, such that all the participant would have to do would be screw-in (twist-in... nuts to Edison...) the lightbulbs and plug the rPi into a port on the back of their router and it would all Just Work.
The lightbulbs would need to be paired to the participants' WiFi, which couldn't be done in advance by the experimenter, but this is a super basic procedure: for Shelly bulbs, the bulb creates a WiFi hotspot when screwed-in to a light socket, you connect your phone/computer to that hotspot and navigate to a given web address where you can put in the password for your home WiFi. Then, you disconnect from the hotspot and the bulb will connect itself to the home WiFi.
Whereupon the Home Assistant rPi, plugged-in to the router, will automatically see it and include it in the program it's running (which the experimenter can set-up identically on all the participants' rPis beforehand). The bulbs only expose themselves to the local WiFi subnet, not to the internet (for absolutely sound security reasons) but if necessary (and if ethically approved) the Home Assistant rPis, which are basically tiny cute little adorable computers, could be exposed to the internet and could report on each bulb's status in real-time to the experimenter (say to monitor protocol compliance, do statistical wizardry with the data or even just to identify and offer tech. support to -or automatically exclude- any participants whose setups somehow weren't working as intended.)
I'm sure you're already in possession of plenty of unsolicited Home Assistant technical advice so I shan't add my voice to the cacophony - but if you do ever want some stolidly-pedagogical, so-simple-it's-almost-patronising guidance or help in setting it up I'd be happy to: feel free to DM me if so.
For the bulb thing, I have some Hue lights which are nice in the bedroom, but they're really expensive and kind of annoying (you need an app to change them). For my bathroom and a few other part of the house, I got lights that change color temperature when you dim them. It's a little finnicky which dimmers they work with and how many you can add to a circuit, but once they're setup it's nice to just slide the dimmer switch at night. The ones I have are pretty expensive but I imagine someone less perfectionist makes a cheaper version with similar features: https://store.yujiintl.com/products/well24-flamewarm-a19-a60-11w-dimmable-led-sleep-bulb-1250k-2pcs-4pcs?variant=45378286584065
Well, I have very mixed feeling about that! On the one hand, I'm incredibly averse to having anything in my home run on apps, and that's a wonderful "midwit-home" solution, exactly what I'd like.
On the other hand, holy shit, $76 for two bulbs?
I actually have some (cheap) bulbs that cycle color temperature when you flick them on and off. I think someone recommended them after my midwit-home post. But I find them so fiddly that I rarely mess with them.
Philips sells a much cheaper version ($2.50 each), but they start out warmer on the cool setting (2700K) and don't get as red at the warm setting (2200K). https://www.homedepot.com/p/Philips-60-Watt-Equivalent-A19-Ultra-Definition-Dimmable-E26-LED-Light-Bulb-EyeComfort-Soft-White-with-Warm-Glow-2700K-4-Pack-576116/321121512
I think other brands exist too, I'm just a light perfectionist.
Just switch bedroom bulbs to smart ones, and set them on a timer to get warmer at night. They are really not that expensive compared to blue-light glasses. And for screens - f.lux will change your life
Yes, I like f.lux also-have used it for a few years on my work computer. Seems to have more capacity than I understand the point of. I keep it set on macular pigment which feels very comforting to my eyes.
One thing that would be easy to do is switch out all bedroom lights to 2700K. I'm thinking of putting a second floor lamp in the room I spend all my time it -- right now I've got a floor lamp with a 5000K LED in there -- but I've got a spare floor floor, so I could just put in a 2700K light in it and switch lamps when/if I think of it. Not ideal, but, still, something.
What's frustrating to me is I don't think 2700K will do *that* much—it seems like it only reduces the fraction of blue light by ~33% compared to 5000K. I want 2000K bulbs!
I went and looked on Amazon, and 2000K LED bulbs are really a thing. They are only equivalent to 60W bulbs (600 Lumens) so pretty dim, but you could plug in 2 or 3. Amazingly enough, they aren't even that expensive -- about $3 a bulb.
https://www.amazon.com/Bonlux-Blocking-Equivalent-Bedroom-Healthy/dp/B0DSHXJNJY/ref=sr_1_8?dib=eyJ2IjoiMSJ9.P9f_7zhLkftAH69QoLFvxAQW67XpAOlKEIrTnF-XjJw0dBPmyX3i8RcTIu4NNHpDh15QzIwAGMX7RzZqlldBWUoCkEob1Rz6iW_mBQEwnmICi5uSa5lRQ03EXkKPb-7Q97vn01E-_FVBAA-_SAlVP4bI_vLX7ry8PcEW-Lw72RCYbzSwihLwAf8Bs4VEcloQUWsdFVIJe5dB9JIAA4VKbVbGS3XVAvX96X-dRUElraEedK4uPUeCZoY0JgHc87xj90YvvIC_4Rf6H6mWzjRNHaHQ726g_QybaQZW-ES3YDI.NjEFrSG6dyqalNyeZzD5yDCrNe4G5HhZc7xjTBlH2TY&dib_tag=se&keywords=2000k%2Blight%2Bbulb&qid=1751582259&refinements=p_n_feature_seventeen_browse-bin%3A9692442011&rnid=9692441011&sr=8-8&th=1
Great post, and I'd love to see a follow up. I recently started using blue light blocking glasses during the day while working on my computer to reduce eye-strain migraines, per the recommendation of an eye doctor. I was extremely skeptical because the sun produces way more blue light than any computer screen. But after trying them, eye strain, which occurred every day very bad, is virtually gone. I understand the detailed physics behind light spectra and am very unsusceptible to the placebo effect. But man, they seem to work. I'd love to see a follow-up to this article where you aggregate these studies about blue light blocking glasses and eye strain. Thank you!
I did stumble past quite a number of papers that looked at stuff like that! I was laser-focused on sleep, but I think there's a lot of research out there...
The part about color perception at the beginning reminded me of this recent paper in which the authors get their subjects to experience a new hyper-saturated blue-green color by firing a laser directly into their M cones: https://www.science.org/doi/10.1126/sciadv.adu1052#
FWIW, I have an orange lamp in my room that I turn on at night (normal bulb, transparent orange paper rolled into a cylinder for the shade). I don't use it for any scientific reasons, I just find it soothing.
Oh wow, that is extremely interesting! The idea seems to be that out three different cones all have somewhat overlapping sensitivity to different wavelengths. So we never (normally) experience just one single type of cone firing. But with this you can. Amazing.
Personally I have 2 sets of lights for day and night. 2500 lux of cool light during the day, then a separate set of Philips hue lights for after work + night.
The hues start as 2700k after work, then switch to red around 9pm.
I'm always sad when I go to a hotel and it doesn't have red lights :(
Some cities have whole districts of rooms with red lights!
In general my midwit-home principles (https://dynomight.net/midwit-home/) make me super hesitant to try set something like this up. This seems like the strongest reason to consider doing smart-home stuff.
Though AFAIK none of the smart bulbs do color temperatures much below 2700K? I want 1800K!
My lifx bulbs go that well and automatically start going lower in colour temp in the evening. This article has inspired me to be more aggressive with that
I had the same scepticism. I started with just controlling everything via hue. Smart lights plus hue smart plugs to control the strings of bright lights. And hue wall switches. It worked well and was zero maintenance.
Eventually I switched to home assistant which is very fun and I have lots of useful automations. But it does completely break once or twice a year. Only worth it if you like tinkering!
The hue-only setup I do recommend though. Zero maintenance it just works