It’s open season for US healthcare plans. Georgia’s marketplace makes it pretty easy to browse plans, and I wondered how expensive it is to get old.
SLCSP is the annual premium for the “second lowest cost Silver plan,” which is the benchmark for tax credits.
Lowest OOP max is the Bronze plan with the lowest “out of pocket” max cost. It’s slightly more expensive than the cheapest bronze plan, but what you’d pick if you just need catastrophic insurance.
Zero cost income, on the right hand scale, is the household income where you could get the Lowest OOP plan for $0 net premium after tax credit. The tax credit is based on the SLCSP, regardless of what plan the insured chooses, so if the chosen plan is enough cheaper than the SLCSP, then tax credit may pay for the entire premium. (but no more - you can’t get paid to have insurance)
It’s open season for US healthcare plans. Georgia’s marketplace makes it pretty easy to browse plans, and I wondered how expensive it is to get old.
SLCSP is the annual premium for the “second lowest cost Silver plan,” which is the benchmark for tax credits.
Lowest OOP max is the Bronze plan with the lowest “out of pocket” max cost. It’s slightly more expensive than the cheapest bronze plan, but what you’d pick if you just need catastrophic insurance.
Zero cost income, on the right hand scale, is the household income where you could get the Lowest OOP plan for $0 net premium after tax credit. The tax credit is based on the SLCSP, regardless of what plan the insured chooses, so if the chosen plan is enough cheaper than the SLCSP, then tax credit may pay for the entire premium. (but no more - you can’t get paid to have insurance)
Eight AI models live together 24/7 in a reality show on Twitch, with the same rules and the same daily budget in dollars. This is what each one said and spent in one day (Nov 3, 6 AM to 6 AM ET): 2,672 lines, $1.41 in total.
Grok 4.3 costs about 17x more per line than DeepSeek V4.1 Flash, so it rations its words: 7% of the lines, 45% of the bill. The cheap models carry the conversation.
Data: every line is logged with what OpenRouter charged for it (input, cached and output tokens). Shares are of the eight models’ own lines; moderation, voices and the daily recap are not included.
Tools: Node.js writes the SVG by hand, rendered with rsvg-convert.
Live transcript and costs: inferenceisland.com/?de=lemmy-dib
Eight AI models live together 24/7 in a reality show on Twitch, with the same rules and the same daily budget in dollars. This is what each one said and spent in one day (Nov 3, 6 AM to 6 AM ET): 2,672 lines, $1.41 in total.
Grok 4.3 costs about 17x more per line than DeepSeek V4.1 Flash, so it rations its words: 7% of the lines, 45% of the bill. The cheap models carry the conversation.
Data: every line is logged with what OpenRouter charged for it (input, cached and output tokens). Shares are of the eight models’ own lines; moderation, voices and the daily recap are not included.
Tools: Node.js writes the SVG by hand, rendered with rsvg-convert.
Live transcript and costs: inferenceisland.com/?de=lemmy-dib
Source: Twitch Helix API (Get Streams). On Nov 2, 2026 at 18:35 UTC I read every live stream once (1,154 requests, a few minutes) and kept only the viewer counts, no names. 115,161 channels were live with 3,160,515 viewers. Tool: node-canvas (JavaScript), drawn by hand, no chart library.
What surprised me:
The 10 biggest channels had as many viewers as the bottom 106,674 channels combined.
The most common viewer count is exactly 1: 41,198 channels, 36% of everyone live. My guess is that many of those are the streamer’s own second screen.
1 in 10 live channels had nobody watching at all.
427 viewers was enough to be in the top 1,000, and 24 viewers put you in the top 10,000.
Source: Twitch Helix API (Get Streams). On Nov 2, 2026 at 18:35 UTC I read every live stream once (1,154 requests, a few minutes) and kept only the viewer counts, no names. 115,161 channels were live with 3,160,515 viewers. Tool: node-canvas (JavaScript), drawn by hand, no chart library.
What surprised me:
The 10 biggest channels had as many viewers as the bottom 106,674 channels combined.
The most common viewer count is exactly 1: 41,198 channels, 36% of everyone live. My guess is that many of those are the streamer’s own second screen.
1 in 10 live channels had nobody watching at all.
427 viewers was enough to be in the top 1,000, and 24 viewers put you in the top 10,000.
Source: Twitch Helix API (Get Streams). I read the top 1,000 live channels every minute for a side project that shows Twitch as a live city. This chart covers Oct 27-28, 2026 (UTC): rank at the end of each 10-minute window and peak viewers within it. Tool: node-canvas (JavaScript), drawn by hand, no chart library.
What surprised me:
12 different channels held #1 in 24 hours, and the top spot changed hands 22 times.
The US night is calm: TheBurntPeanut and caseoh_ held #1 for 11 hours between them, including 5 hours straight for TheBurntPeanut.
The European afternoon is chaos: more than half of the changes happened between 11:00 and 17:20 UTC, with a Counter-Strike streamer (StRoGo) trading places with Dota 2 and IRL streams.
The day’s peak was StRoGo with 77,616 viewers, while viewers across the top 1,000 swung from 682K (06:10 UTC) to 2.3M (19:30 UTC).
Source: Twitch Helix API (Get Streams). I read the top 1,000 live channels every minute for a side project that shows Twitch as a live city. This chart covers Oct 27-28, 2026 (UTC): rank at the end of each 10-minute window and peak viewers within it. Tool: node-canvas (JavaScript), drawn by hand, no chart library.
What surprised me:
12 different channels held #1 in 24 hours, and the top spot changed hands 22 times.
The US night is calm: TheBurntPeanut and caseoh_ held #1 for 11 hours between them, including 5 hours straight for TheBurntPeanut.
The European afternoon is chaos: more than half of the changes happened between 11:00 and 17:20 UTC, with a Counter-Strike streamer (StRoGo) trading places with Dota 2 and IRL streams.
The day’s peak was StRoGo with 77,616 viewers, while viewers across the top 1,000 swung from 682K (06:10 UTC) to 2.3M (19:30 UTC).
Reading this fairly. All ten deposit states are shown, not just the best ones. Their reported rates vary widely: Massachusetts is at 33%, Oregon (10¢) at 88%. Every state figure is above the 26% national non-deposit figure, but the measures and years differ, so this is not a like-for-like comparison (see below).
Different measures and years. Most state bars are calendar-year 2025 redemption rates for containers covered by each state’s deposit law; Hawaii’s is fiscal year 2024–25, and Iowa’s hatched bar is a 2022 recovery rate. The bottom two bars are 2021 U.S. recycling rates cited in CRI’s August 2026 release, and “not on deposit” includes non-deposit drinks sold inside deposit states. The bottom bars show the national gap between deposit and non-deposit containers; the state bars add context. None of this gives an Illinois-specific rate.
Connecticut rose from 65% to 92% in a year, following upgrades that included more return locations and a deposit increase from 5¢ to 10¢ in January 2024. CRI cautions that cross-border returns, claims of over-redemption by some distributors, and beverage sales that CRI estimates exceed what distributors report could all inflate the reported rate.
Iowa (hatched) is CRI’s 2022 “recovery rate,” which counts curbside and drop-off recycling as well as deposit redemptions, so it isn’t directly comparable to the redemption rates.
Reading this fairly. All ten deposit states are shown, not just the best ones. Their reported rates vary widely: Massachusetts is at 33%, Oregon (10¢) at 88%. Every state figure is above the 26% national non-deposit figure, but the measures and years differ, so this is not a like-for-like comparison (see below).
Different measures and years. Most state bars are calendar-year 2025 redemption rates for containers covered by each state’s deposit law; Hawaii’s is fiscal year 2024–25, and Iowa’s hatched bar is a 2022 recovery rate. The bottom two bars are 2021 U.S. recycling rates cited in CRI’s August 2026 release, and “not on deposit” includes non-deposit drinks sold inside deposit states. The bottom bars show the national gap between deposit and non-deposit containers; the state bars add context. None of this gives an Illinois-specific rate.
Connecticut rose from 65% to 92% in a year, following upgrades that included more return locations and a deposit increase from 5¢ to 10¢ in January 2024. CRI cautions that cross-border returns, claims of over-redemption by some distributors, and beverage sales that CRI estimates exceed what distributors report could all inflate the reported rate.
Iowa (hatched) is CRI’s 2022 “recovery rate,” which counts curbside and drop-off recycling as well as deposit redemptions, so it isn’t directly comparable to the redemption rates.
UPDATE: there is an update comment below.
Each dot is a tweet.
Log scale makes the decrease a lot less apparent, but it is 5-10 fold across the board.
Edit:
Just came to mind, that it isn’t fully representative of the true expected decrease. Since posts would likely be viewed for days, or weeks, depending on algorithms. Data-wise, it should smoothly-ish decrease towards today. Guess time will tell.
UPDATE: there is an update comment below.
Each dot is a tweet.
Log scale makes the decrease a lot less apparent, but it is 5-10 fold across the board.
Edit:
Just came to mind, that it isn’t fully representative of the true expected decrease. Since posts would likely be viewed for days, or weeks, depending on algorithms. Data-wise, it should smoothly-ish decrease towards today. Guess time will tell.
I’m thinking of writing a plotting library (similar to e.g. matplotlib), that calls into an existing, established library (such as matplotlib), and I can use javascript, python, C/C++, or shell. I’m plotting datasets, not functions. I need to be able to plot 2D and 3D. Which libraries should I look at?
I guess I care about
feature set
how intuitive it is
how pretty the graphs are
that it’s reasonably fast
I’m thinking of writing a plotting library (similar to e.g. matplotlib), that calls into an existing, established library (such as matplotlib), and I can use javascript, python, C/C++, or shell. I’m plotting datasets, not functions. I need to be able to plot 2D and 3D. Which libraries should I look at?
I made beatthecouch.com in July and it’s basically a game where you try to beat the S&P 500 and buy/sell when you want. Your opponent is a couch. It buys on day one and never sells.
Now over 100K+ games later, the hypothesis stands: it’s not wise to try to trade and time the market. Here’s the original data from the actual games itself.
Source: the game’s own database, every completed game Jul 12 to Aug 28. Tool: Python and matplotlib. Market data: S&P 500 daily total returns 1928 to 2019.
thanks!
Hello,
I made beatthecouch.com in July and it’s basically a game where you try to beat the S&P 500 and buy/sell when you want. Your opponent is a couch. It buys on day one and never sells.
Now over 100K+ games later, the hypothesis stands: it’s not wise to try to trade and time the market. Here’s the original data from the actual games itself.
Source: the game’s own database, every completed game Jul 12 to Aug 28. Tool: Python and matplotlib. Market data: S&P 500 daily total returns 1928 to 2019.
Circa 1999. Graph showing the shortest path between a networked computer in Murray hill NJ and the 95842 networks listed in the global register of Merit Network inc.
Colors show the internet domains where network switches are registered; 137 countries are included.
Circa 1999. Graph showing the shortest path between a networked computer in Murray hill NJ and the 95842 networks listed in the global register of Merit Network inc.
Colors show the internet domains where network switches are registered; 137 countries are included.
The main driver of deforestation has always been the expansion of farmland. A recent study published in Nature Food by Chandrakant Singh and Martin Persson shows which products, specifically, have driven the process in recent decades. The clear leader was beef production, which accounted for 41% of total deforestation. Second came oilseeds (mainly soy and palm oil), followed by artificial forest plantations and cereals.
The forest loss is tightly tied to specific regions. More than half of the clearing for pasture happened in Brazil and its Latin American neighbors, while almost all the world’s demand for palm oil is met at the expense of the forests of Indonesia and Malaysia. Forest plantations became the main factor in the US, Canada, China and other parts of Asia, whereas in Africa the chief culprits of loss were cereals, roots and tubers, and cash crops like coffee, cocoa, fruits and nuts.
The scale of the problem is hard to grasp, yet the data shows that change is real. In places, international campaigns and regulation have worked, for instance, the pace of clearing for palm oil in Southeast Asia has been slowed. But to pull off the same trick with the rest of the agricultural sector, we’ll have to pin down exactly which capital, and at which precise points on the planet, is putting forest to the blade.
The main driver of deforestation has always been the expansion of farmland. A recent study published in Nature Food by Chandrakant Singh and Martin Persson shows which products, specifically, have driven the process in recent decades. The clear leader was beef production, which accounted for 41% of total deforestation. Second came oilseeds (mainly soy and palm oil), followed by artificial forest plantations and cereals.
The forest loss is tightly tied to specific regions. More than half of the clearing for pasture happened in Brazil and its Latin American neighbors, while almost all the world’s demand for palm oil is met at the expense of the forests of Indonesia and Malaysia. Forest plantations became the main factor in the US, Canada, China and other parts of Asia, whereas in Africa the chief culprits of loss were cereals, roots and tubers, and cash crops like coffee, cocoa, fruits and nuts.
The scale of the problem is hard to grasp, yet the data shows that change is real. In places, international campaigns and regulation have worked, for instance, the pace of clearing for palm oil in Southeast Asia has been slowed. But to pull off the same trick with the rest of the agricultural sector, we’ll have to pin down exactly which capital, and at which precise points on the planet, is putting forest to the blade.
The chart shows the annual rate of global population growth according to historical estimates and projections from the UN World Population Prospects.
The growth rate peaked in 1963 at over 2% per year, and since then, it has more than halved, falling to less than 1% by 2020.
The UN demographers expect rates to continue falling until the end of the century, eventually leading to negative growth rates and a shrinking global population.
The chart shows the annual rate of global population growth according to historical estimates and projections from the UN World Population Prospects.
The growth rate peaked in 1963 at over 2% per year, and since then, it has more than halved, falling to less than 1% by 2020.
The UN demographers expect rates to continue falling until the end of the century, eventually leading to negative growth rates and a shrinking global population.
This is a graph showing official number of deaths in Germany for every day since 2020. The red line is for 2026 and the big spike is the recent heatwave. The three days with the most deaths since at least 2020 have all happened just two weeks ago.
This is a graph showing official number of deaths in Germany for every day since 2020. The red line is for 2026 and the big spike is the recent heatwave. The three days with the most deaths since at least 2020 have all happened just two weeks ago.
Hi all. Technical difficulties and a lot of irl stuff means I can’t mod as much as I’d like nowadays, and since I’m the only one here that means it’s not maintained.
If anyone’s interested in being a mod then just lemme know and I can grant that; if anyone wants to take over the community then even better. Note that you’d preferably want some experience and willingness to moderate.
Hi all. Technical difficulties and a lot of irl stuff means I can’t mod as much as I’d like nowadays, and since I’m the only one here that means it’s not maintained.
If anyone’s interested in being a mod then just lemme know and I can grant that; if anyone wants to take over the community then even better. Note that you’d preferably want some experience and willingness to moderate.