doubleunplussed (@doubleunplussed) header
doubleunplussed (@doubleunplussed) profile picture

Atomic physicist and embedded software engineer, general-purpose nerd. Forecasting, Linux, running, lifting, psephology, economics, Georgism, other nerdery

Melbourne, Victoria
Joined February 2023
Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)Photo by doubleunplussed (@doubleunplussed)
if you missed the slightly insane AI safety schizopoast, I put it into a song for you and you should turn the sound on right now because we're so fucking back
134
252
73
1,925
498,077
doubleunplussed retweeted
Throwing rubbish in a landfill is (usually) better for the environment than recycling it. With few exceptions, recycling ends up emitting more CO2 and costing more money, while landfills are cheap and secure. - Recycling mid-tier plastics (like plastic bottles or butter tubs) costs thousands of dollars per ton of CO2 saved. It's cheaper to do direct air CO2 capture than that! And there is no shortage of plastic, nor any problem with sequestering it in the ground once we're done with it. - Modern landfills are extraordinarily secure and are *much* more than just big piles of trash. They capture 75%–95% of methane gas. Groundwater contamination is doesn't really happen with landfills built since 1990. A study of 31 landfills built since 1988 in NY found no adverse effect at all on local groundwater quality. - The crusade against single-use plastics has driven us to products that use far more materials and energy and would need to be used an absurd number of times to actually be an improvement. A steel straw needs to be used 150 times to be worth using. A ceramic mug needs to be used 1,000 times to beat a polystyrene cup. A cotton bag needs to be reused 173 times to beat single-use. - Some things ARE worth recycling. Recycling metal saves a huge amount of energy and material – recycling aluminium saves 95% of the energy spent on making fresh stuff. So we do a lot of it already: about 30% of steel worldwide is recycled. Modern, well-run landfills cause minimal environmental damage. We have plenty of space for them – less than 0.01% of the US is used for landfill and only about 0.02% of Germany. We can even build on top of old landfills, so the land area is not spent permanently. New in @WorksInProgMag, @robertwiblin and @chalmermagne on why it's usually best to just bury your trash. worksinprogress.co/issue/jus…
268
789
285
5,476
1,619,842
doubleunplussed retweeted
Replying to @xlr8harder
Oh shit, wonder if it's just increasing neuroplasticity. Like your brain is still messed up from Alzheimer's but if we shuffle things around a bit maybe we can get it kinda working for a bit longer. (if it works at all - my default assumption is things like this don't replicate)
12
10
1
2,244
71,353
I conceive of "social surface area that can lead to marriages" as a public good that we have mismanaged. If two people meet at a coffee shop and get married, the couple benefits enormously, the community benefits somewhat, and the coffee shop just gets $10 for 2 lattes.
8
7
2
200
5,837
Blue Origin's New Glenn just blew up at LC-36 while attempting to Static Fire ahead of NG-4. nsf.live/spacecoast
1,462
4,966
2,725
20,239
8,040,746
(Georgist meeting) we recognize that we are holding this meeting on untaxed land,
19
726
13
7,581
171,652
doubleunplussed retweeted
I think what the site really needs is to match up new joiners (or existing ones with low followings) with each other to jump-start a bit of a network for them. Otherwise you just end up shouting into the void or replying to accounts with huge followings that won’t engage back.
Over the last few months, we scoured the world for the top posters in every niche & country We've compiled them into a new tool called Starterpacks: to help new users find the best accounts—big or small—for their interests ⬇️ Reply below with a topic you're most interested in We'll be rolling out to everyone in the coming weeks.
15
2
1
110
15,321
doubleunplussed retweeted
We estimate that, on our tasks, Claude Opus 4.5 has a 50%-time horizon of around 4 hrs 49 mins (95% confidence interval of 1 hr 49 mins to 20 hrs 25 mins). While we're still working through evaluations for other recent models, this is our highest published time horizon to date.
64
263
200
1,990
1,354,922
The people always tell me that I can't fit a sigmoid to just one data point but they don't understand that I'm a WORLD CHAMPION FORECASTER and that means I can do what I want
Peter, the more relevant METR data would probably be % task completion over time, not time horizon. But generally I think it's too early to be fitting a sigmoid to RLI.
3
9
1
117
14,081
Replying to @emilykmay
I have some optimisation obsessions, and think it's fine as long as you know it's a hobby Like it can overlap with actual life improvement, but the obsessive part is a hobby, which is fine as long as you don't mistake it for necessity or judge others for not doing same
1
123
doubleunplussed retweeted
I did a deep dive into this CR investigation for my blog. Their reporting has a number of concerning errors that completely undermine their findings. Chief among these is a reliance on CA's pseudo-scientific daily lead limits that are far lower than average human intake.
my latest investigation for @ConsumerReports is based on months of reporting and 60+ lab tests of leading protein supplements we found that most protein powders and shakes have more lead in one serving than our experts say is safe to have in a day — some by more than 10 times !
19
32
24
349
130,529
Say it from the rooftops "If you are spreading negative emotional contagion, there is a very good chance that, no matter what you are saying, that you are part of the problem.  A more fundamental division these days than Left vs. Right." - @tylercowen
4
16
4
124
19,738
doubleunplussed retweeted
look, it's not fun to live in a country where a hundred million people don't want us to have the same rights to live in peace and start a family that they enjoy. it sucks. it is painful and alienating and scary and unfair. Now it's escalating in a way that goes way past 'unfair' to powerful people calling for horrific acts of violence. I guarantee you I am speaking from first hand experience here and that yesterday my dashboard was full of people calling for the murder of people in my immediate family. No one ever has an obligation to argue with people who hate you. It's fine to block them all. But if you want anything to change, you have to succeed at persuasion. Over the last ten years I think there was hope that this wasn't true, that all reasonable people had already agreed and at this point the argument was over and we could just tell people that the questions were settled and they were bigots if they didn't still agree. But this didn't...work. It didn't make them kind. It didn't make them accepting. It just made them resentful, led to backlash, and made everyone retreat to media bubbles where they only had to hear what they already believed. And frankly I think it contributed to a nihilistic younger generation which doesn't believe in persuasion. You don't have to argue about this stuff. But someone has to. Ezra is right that it'd be good if there were a Charlie Kirk of the left, because persuasion really, really matters.
6
3
91
3,040
doubleunplussed retweeted
There are a bunch of really basic and easy ways to improve your social media experience that I see smart people not doing. 1. Turn off auto-playing wherever possible 2. When you see something that you would prefer not to have seen, consider why it's on your feed, and use the tools to remove it. You can unfollow people, mute people, mute words, or turn off retweets from people 3. Deliberately don't engage with things you want to see less of. If you engage with things because they make you angry or scared, social media will dump more of those things on you. Engage with what you want to see more of
16
24
3
275
15,249
If we did this, people would quickly start to measure daily caloric expenditure in grams ("basal dry mass expenditure rate", "sweat-free exercise mass expenditure rate"), and multiply the weight of food by a food-specific "food mass retention factor" for how much of the mass can be expected to take part in the calculus, and we would quickly find we were simply doing calories in calories out but in units of grams of body fat (1g=7.7 kCal) instead of kCal. Calories is a useful level of abstraction. I understand it clearly doesn't work well (meaning: it's very difficult/not worth it to be in caloric deficit) in many circumstances, presumably due to hormonal differences between people, or hormonal changes caused by weight gain itself, that we are yet to fully understand. Nonetheless I think calories is the right level of abstraction to think about weight change for most people. People doing weight training bulk and cut regularly with good success rates, by counting calories. In these circles calories-in calories-out is not even a question, it's just fact that that's what you have to do, confirmed repeatedly by experience. I'm losing weight now by counting calories, it's working. I've done it before and I'll do it again. I'll continue to recommend it to others. I understand it won't work for everyone, but I think the NYT is doing people a disservice by suggesting they dismiss it. It should be plan A for most people. (note: lots of water weight change in this plot than can't be attributed to fat - but ignore the first four and the last datapoint and the trend is there. I predict it will continue at about 0.5kg/w for as long as I maintain this deficit.)
1
1
12
397
Replying to @robertwiblin
For me with software work the transition was Claude 3.5 Sonnet (June 2024), when LLMs went from being a toy, or something with narrow use cases, to being something I reached for every day. Before that I did not interact with an LLM most workdays. The original 3.5 Sonnet I think mostly useful for work in a language/domain that was new to me, then with 3.5 Sonnet "new" (Oct 2024) I started to reach for Claude even in many contexts where I'm more or less an expert already. Around about o3-mini (Jan 2025) was when they became practically useful for symbolic maths for me. I basically always start a literature search using ChatGTP's deep research (Feb 2025) these days. Since Gemini 2.5 Pro (March 2025) I no longer read textbooks alone - I upload a PDF to Gemini and have it tell me which bits to read given what I want to know, have it critique my understanding etc. These days if I want to understand a codebase I run Claude Code (June 2025) in a clone of it and ask it what I want to know. Today Claude Code translated a small Go library to Python for me with only modest hand-holding. Still not using AI coding tools much for "serious" coding though, it requires too much hand-holding to usually be worth letting them actually write the code. So we're not at that threshold yet for me - I use AI to learn and give advice, but am still fairly tightly holding the reins unless it's a one-off script I don't care much about. Feels like a series of phase transitions even though METR's charts show a continuous smooth curve
1
2
210
I just started reading an absurdly dumb take from this site out loud, and @MaskedTorah was like "Peter, you can't let yourself feel reward signals from this. You just have to hit 'see less of this' and move on immediately.". I think that should be much more my policy now.
7
8
2
193
12,310
Not relevant to this situation I think, but I find pure freedom of association to be a bit self-defeating without some guardrails in the form of anti-discrimination laws and the like. Left to their own devices, people and organisations end up using their freedom of association to coerce others regarding who they associate with, and you can get a purity spiral or other race to the bottom where freedom of association ends up worse for everyone in practice. I'm not wed enough to the pure idea of freedom of association to be comfortable with that outcome as a natural consequence of it, so I think restricting freedom to some extent can make sense in order to prevent it being used to enact its own demise. (Acknowledging that we should be extremely suspicious of appeals to restricting freedoms for the greater good, which are very often made naively or in bad faith. Pushing back on such appeals is a good heuristic, but that doesn't mean there aren't cases where the heuristic misfires, and I tentatively think this is one of them) Maybe it would be pretty much fine if the norms/laws were (somehow) only about higher-order coercion - you can always dissociate from anyone you like, but if you're dissociating because of who *they* associate with, that's frowned upon, or illegal in an employment context - though actual laws are hard to get right so I won't pretend I know how to implement that sanely. Again, not relevant I think to the above situation which seems pretty close to a pure case of someone exercising freedom of association, and not using it coercively or under coercion.
1
77