There are definitely less readable blogs, even restricting to ones that aren't intentionally hard to read. For example: https://www.lilywise.com/amusement (Disclosure: written by my kid, who was just shy of 7yo then)
Personally, I like Gwern's style and aesthetic a lot, and don't have trouble reading his stuff.
It's pretty unfortunate that the Yudkowsky-and-LessWrong crowd picked a term that traditionally meant something so different. This has been confusing people since at least 2011.
I'm really glad that you're making and selling far-UVC products. Nukit is the other manufacturer (aside from Aerolamp) that I feature on https://www.faruvc.org
I really hope that your thesis is correct, and we end up with widespread low cost high quality unencumbered far-UVC. In my looking, though, it seems like bulb life is an issue? Is that right, or have I been snookered by Ushio's marketing?
Well, one reason KrCl lamp life became an issue is that they are not replaceable in any commercially available filtered Far-UVC fixture. Currently, at the end of the KrCl excimer lamp life- typically 3000-5000 hours, you throw the entire Far-UVC device away- filters, ballast, control electronics- the whole thing in the trash. There is some talk of possible refurbishment, but at high cost with no companies really in place offering it.
There's really no excuse for this but planned obsolescence. Replaceable light bulbs are 140-year-old legacy tech and trivial to engineer.
Needless to say, this was a major stumbling block for institutional buyers- having to buy and reinstall ~$5000 worth of Far-UVC for a small room every year or two. At which point virtually every Far-UVC company started claiming everything from 5,000 hours to 20,000 hours operating life- with absolutely no data to support this. Some cite in-house testing, but no real evidence is offered.
Given that fraudulent claims in the Far-UVC industry are common, there's every reason not to trust marketing claims that aren't backed by third-party tests. If you don't have a test that proves it, don't say it. (We urge our customers to trust no one, not even us. "Trust" has no place with potentially dangerous UV devices- everything must be third-party lab tested).
A KrCl excimer lamp that could exceed the current, widely documented 3,000-5,000hr (30% degradation or L70) operating life would be huge news and we'd love to see it. But there would be patents, papers, independent lab tests, some paper trail other than manufacturer claims to support this.
Right now, it's all "trust me bro" marketing with perhaps bit of hedging about possibly using reduced power or limited operating time to increase lamp life. By the time buyers have reached 5,000 hours, the products are well out of warranty, and they are left without recourse. If someone says anything about their Far-UVC lasting more than 5,000 hours, the correct response is "Citation?".
In the absence of any credible data showing that anyone is getting an L70 over 5,000 hours of use in the real world at 100% power and uptime, we've chosen to focus on low-cost, non-IP-encumbered, replaceable bulbs, and all of our future products will use them. Any other manufacturer is welcome to use them as well, with the hope that this will get Far-UVC ownership down to the "rice cooker or oscillating fan" price range needed for those in the most desperate need. Because until everyone is protected, no one is. We're all just a cough and a plane ride from each other.
I have never seen any other data for any other commercially available krypton-chloride excimer lamp besides this paper and this poster. All that is "widely documented" is manufacturers claims. The non-USHIO KrCl excimer lamps I have personally tested had L70s closer to 1000 hours, nowhere near 3000-5000 hours. Of course since I am involved in Aerolamp I am not a credible third party, but I am not asking anyone to believe me, I am simply explaining why I decline to believe this 3000-5000 hour claim that is presently without any substantiation whatsoever, not even a paper or conference poster.
Of course, a cheap 1000-hour bulb that is easy to replace and available from multiple manufacturers is great! For many people this approach makes sense. But it is not a 3000-5000 hour bulb and consumers need to know that in order to replace their bulbs in a timely fashion.
You will probably say that the paper and poster and is not credible third party data. This is why Aerolamp has submitted multiple samples to LightLab Allentown for lifetime testing. Perhaps you would be willing to do the same, given how frequently you emphasize the importance of third-party data?
The scientists who wrote this paper work directly for Ushio, correct? And it has not been replicated or verified in any way?
>I am simply explaining why I decline to believe this 3000-5000 hour claim that is presently without any substantiation whatsoever
Very sensible, it should be checked also and claims until then retracted.
>This is why Aerolamp has submitted multiple samples to LightLab Allentown for lifetime testing. Perhaps you would be willing to do the same, given how frequently you emphasize the importance of third-party data?
Sounds like a great idea, we have multiple new gas formulations to test and low-cost replaceable bulbs mean it will be easy to offer customers greater operating life as we improve them. Perhaps Aerolamp would also like to refrain from making any claims without third party data as well? It would be a good example for us to set.
Sure, I'm game! I can loop you into our comms with LLA and you can do the same, and they can send us both the results as they come in. OSLUV says they're happy to pay for it too so no worries about the expense
I really do think the ushio bulb will last 13.5k+ hours though, and that's a long time to wait--how about we publish the LLA results at 1000 hours, and every subsequent 1000 hour interval after that? At minimum, it should give us a sense for how the lamps are performing, even though we might not get the L70 lifetime for a while. Then, as soon as LLA has confirmed receipt of both our samples we can both retract our lifetime claims, at least until the first 1000 hour interval
Just let me know a good email for you and we can get the ball rolling on this, you can get me at vivian@aerolamp.net
This isn't the kind of thing you can do with a license, as long as training a model doesn't require a license. Now, that's an open question legally in the US, and there are active lawsuits, but that does seem like the way it's most likely to play out.
Why do you think there's was an implicit agreement that documentation was only intended for humans? I've written a lot of documentation, much of it open source, and I'm generally very excited that it has proved additionally useful via LLMs. If you had asked me in 2010 whether that was something I intended in writing docs I'm pretty sure I would have said something like "that's science fiction, but sure".
You still intended it for humans. Intent is defined by what one is aiming for, and without knowledge of an alternative, that was your intent.
100% I get that you are OK with it being used by non-human ingestion. And I think many might be OK with that.
One thing, I'm not sure how helpful the documentation is. I think we're getting training out of example, not docs. This makes me think... we could test this by creating a new pseudo-language, and then provide no examples, only docs.
If the LLM can then code effectively after reading the docs, we'd have a successful test. Otherwise? It's all parroting.
This absolutely depends on the frequency of UVC and the intensity of the lamp. The lamps this post links such as https://aerolamp.net are putting out 222-nm, which is much safer than longer UVC wavelengths and the intensity is well under TLV when placed 8.5ft up (or higher).
They can leak into higher wavelengths. You are really putting a whole lot of trust in manufacturers if you are sitting underneath one of these for decades with unprotected eyes. Not a risk I would take personally (I have glaucoma already, so I'm a bit more sensitive than the average person about eye health)
Kinda like advertising "Asbestos-Free Cereal" isn't it? If someone was marketing a product to me and they were super insistent about how super duper safe it was I would probably start getting suspicious
UV rightfully raises concerns about skin damage, highlighting that they're careful about excluding the harmful parts would be helpful for customers who either know just enough to think "UV bad" or to those who wonder how narrow their filters are.
Imo a better analogy would be selling a circular saw with a safety mechanism and hiding the latter in the specsheet.
Personally, I like Gwern's style and aesthetic a lot, and don't have trouble reading his stuff.
reply