To me, this looks like a technical mistake rather than an intentional leak.
This was my immediate thought.
Over at SCOTUSBlog today, one of the commenters asked if the clerks were working 24-hour days at this point. My thought on that was no, probably not, but 18 hours, sure. When I worked for the legislative budget committee in my state, the end of the session was insane. The last crush of bills, and amendments, and while all of those were in a management system, a single mistyped date/time or wrong key press could release something too soon. And staff was always short on sleep the week leading up to it.
Margin of error today depends on knowledge (or guesses) about the underlying distribution. We all know that somewhere out there a lead researcher and minions graduate students are fitting a large neural network to the massive data sets for voter registration, fine-grained economics, global ammunition consumption, etc. Probably there's more than one. They won't be players this year, but at least one will get lucky with choice of NN structure and data sets and will be doing a bang-up job of forecasting by 2028. No one knows what the distribution behind the NN outputs is. No one is going to have a handle on margin of error.
The high school football kids who are going to be big-time college players go to training camps in the summer, often by position. Where better coaching staffs than they have at home are going to correct their technique, change their exercise programs, hammer on their knowledge of the game.
The game's too violent to do full-contact stuff at the camps.
Are you not the same LeeEsq that comments at LG&M? Have you been ignoring Denverite's stories there about his basketball-playing daughter for the last few years? School team. Local club team. Travel in the summer.
ThTh3: I assume that the generative forms of large neural net models will be useful for things eventually. I'm much more interested in the recognition sorts of things that are already useful. For example, significant improvements have been made in optical character recognition accuracy. And stuff that can be built on top of high-quality recognizers.
I have always been a very warm sleeper. In the summer, most times even a top sheet is too much, I shove all the covers aside and sleep without any of them. In the winter, a top sheet and single blanket is usually enough. Of course, if I'm sharing the bed in the winter -- spouse, pet, whatever -- at some point I wake up because I want to roll over but can't, there's someone snuggled right up against me to stay warm.
IIRC, Congress passed specific legislation some time back stating that members of the armed forces may bring their non-citizen spouses into the country.
Not for years. Next step is probably to appoint a new special master, since the current one didn't get the states to come to a proper decision. This will be either the fourth or fifth SM for the case, I believe.
From memory, so suspect, but the US objection is that the consent decree didn't settle the issue of whether withdrawals from an aquifer that is hydrologically linked to a surface water flow is a withdrawal from the surface water. The engineering answer to the question has always been yes, with a multiplier between 0 and 1. Most western state water law has always assumed the answer is no. If the Supreme Court ever holds that the engineering answer is the correct one, all hell will break loose.
Twenty extended families in the Imperial Valley get more water from the Colorado River than Nevada's entire allocation. The families do a lot of vegetables, but their primary cash crop is alfalfa.
My father came out of his stint in the US Navy a senior petty officer. Nothing was ever quite clean enough for him, nor we kids' beds made properly. When my wife-to-be and I moved in together, I carefully refrained from remaking the bed after she had done it.
Hence the common start-up practice of paying crap, with limited benefits, jammed into minimum space, but with a pile of shares and/or options. Intel millionaires. Microsoft millionaires. Lambda School was almost certainly spending the money differently than on generous salary/benefits for <200 employees.
Nvidia's book value is somewhere over $3 trillion right now. How many of the early people are cashing in and leaving?
HOW IN THE HELL DO YOU BURN THROUGH $120 MILLION DOLLARS?
Many years ago -- early 1990s -- I was working at a giant telecom company. The company created a project to build a new billing system without some of the serious limitations of the old system. The project burned through $150 million dollars and produced not a single line of usable code. The company reorganized the technology part of the business for the sole purpose of putting a particular SVP and a particular 200 people into a single organization, fired them all, then three months later reorganized back into something sane.
If you're looking for specific methods, consultants are good. As I recall, the group of consultants that did an incredibly obtuse object-oriented design got a million dollars. When a couple of us who were far enough away from the project that we could safely ask questions asked how big the server farm would be to use the architecture to handle 14 million customers, they said they didn't know but for two million dollars and another nine months they could tell us. Buying hardware because the schedule says it's time to buy the hardware even though there's no actual software architecture is very good.
ETA: Misthreaded. This should be attached to JB's 5:43 pm comment.
JT8: I listened to this when it first went up, and the question I was left with was "How many people does it take to go from being an outpost to a colony?" Bases in Antarctica and the ISS are clearly outposts. Basically, they make no effort to become self-sufficient or to rise to the level of technology required to put them in place. Lots of people talk about colonizing the Moon; Elon Musk talks about colonizing Mars; how many people?
The question gets asked, indirectly at least, in a variety of speculative fiction. Heinlein's lunar colonists in The Moon is a Harsh Mistress clearly number in the millions. They're still not self-sufficient. Two examples: manned spacecraft and big computers are come from Earth, not local producers. James Blish's Cities in Flight books imply millions in a couple of different ways. Small planetary colonies drop back to lower technology levels, creating markets for the migrant cities' technology. Even hundreds of cities with populations totaling millions struggle. Exchange of specialists and technology between the cities is a regular occurrence.
My own guess is that the minimum population required to maintain our current level of technology is around 30-40 million. Very few of those people will be involved with the leading edge of technology. Far more of them will be farmers, teachers, nurses, bartenders, etc.
Based purely on muscle motions, I'd speculate that they pulled the trigger once. Then they held the finger as still and rigid as they could while the trigger banged into it 19 times.
ETA: This is why gun safety courses teach you not to put your finger inside the trigger guard until you are ready to shoot. Because if you just leave the finger there, sometimes the trigger bangs into it and the weapon fires.
The wording in the forecast discussion from my local NWS office is almost always a little amusing. Altitude of the area they cover ranges from 4,500 ft to 14,000 ft. In some events, if the storm track is wrong by 50 miles precipitation goes from 4" of snow to 24". A very small error in the altitude of the moisture-bearing layer is the difference between virga and that 0.01" of rain.
When I worked on the U of Colorado campus in Boulder, one of the (probably) academic myth stories that got told was about an intro to meteorology class where students had to turn in their forecast for the next day each time the class met. Some wag supposedly turned in the NWS forecast, and finished next-to-last in the class.
I have read lately that some of the deep machine learning people are building billion-coefficient models for doing forecasts in difficult areas. I would not be surprised if those are more accurate for my local area than the physics-based models.
I'm backing the low-probability outcome on this one -- they schedule a rehearing for next term, and put the decision off until after the November election results are known.
That's a reasonable thought. Also, given the reaction to Dobbs, five months before the presidential election is not a good time to do a nationwide ban on the most common method of abortion if you want to see more Republicans elected.
That would be, at best, term limits in another form. I worked for the Colorado state legislature about the time the state's term limits were starting to really bite. Every two years, after elections, 25-30% of the members were serving their first term. Staff ran training sessions. I was on the budget staff and it was clear from the questions that the new members had absolutely no idea about how the state budget process worked, nor all the constraints on it. It was clear that when the budget staff director was making assignments that he took preservation of institutional memory as a concern, since that memory was no longer going to reside with the members.
Being a father isn't something you get to give up when your children reach a specific age. Nor, based on my own experience, is being someone's child something you get to give up. When I had to put my wife in memory care a bit over a year ago, my children's reactions, both in their own way, were "Shut up, Dad, and let us help."
I think 538's polling model is missing an important point. Or perhaps the pollsters' models are missing it. From 1988, the Democratic share of the presidential vote in CA, CO, AZ, and NV has increased by between one and two percentage points per four-year cycle. R-squared values from 0.6 to 0.9. I expect Arizona and Nevada will go for Biden where 538's model favors Trump narrowly in those two.
I'm starting a new applied math project and am in the process of skimming technical journal papers, and reading a subset in detail. The non-political part is "Old mathematicians don't retire, they just go amateur again." The political part, about which I should do a separate post, is that an amateur mathematician on a retiree's budget has to steal the papers via the Russians.
AAA has for years offered a buying service, at least in some states. A number of companies advertise that you pick the car and the listed price from their web site and they'll deliver it. Some will also make the deal for your trade-in, in which case the new car comes on a flatbed and they take away the old one.
I had been thinking about replacing my Honda Fit before we had to put my wife in memory care. That costs as much as a modest new car per year, even net of long-term care insurance, so I've reconsidered. Something electric. But no more than I drive, I probably can't even justify it on carbon footprint. New car carbon cost, plus local power won't be carbon-free for several more years vs. my modest (by US standards) gasoline usage. The Fit will be 16 in August. I make it a 50/50 bet whether I have to give up driving before it dies.
I did not, and do not now in syndication, watch Big Bang Theory. Summarize it as "These are my people." And I don't care to watch them made fun of.
Of course, I'm the person who was asked "Who are your people, Mike?" at a party where I wasn't paying much attention to the conversation, and answered "The applied mathematicians."
*Comment archive for non-registered commenters assembled by email address as provided.
On “Not Again: SCOTUS Has Another Abortion Ruling Leaked Early, Read It For Yourself”
To me, this looks like a technical mistake rather than an intentional leak.
This was my immediate thought.
Over at SCOTUSBlog today, one of the commenters asked if the clerks were working 24-hour days at this point. My thought on that was no, probably not, but 18 hours, sure. When I worked for the legislative budget committee in my state, the end of the session was insane. The last crush of bills, and amendments, and while all of those were in a management system, a single mistyped date/time or wrong key press could release something too soon. And staff was always short on sleep the week leading up to it.
On “Has The Election Turned A Corner?”
Margin of error today depends on knowledge (or guesses) about the underlying distribution. We all know that somewhere out there a lead researcher and
minionsgraduate students are fitting a large neural network to the massive data sets for voter registration, fine-grained economics, global ammunition consumption, etc. Probably there's more than one. They won't be players this year, but at least one will get lucky with choice of NN structure and data sets and will be doing a bang-up job of forecasting by 2028. No one knows what the distribution behind the NN outputs is. No one is going to have a handle on margin of error.On “Open Mic for the week of 6/24/2024”
The high school football kids who are going to be big-time college players go to training camps in the summer, often by position. Where better coaching staffs than they have at home are going to correct their technique, change their exercise programs, hammer on their knowledge of the game.
The game's too violent to do full-contact stuff at the camps.
"
Are you not the same LeeEsq that comments at LG&M? Have you been ignoring Denverite's stories there about his basketball-playing daughter for the last few years? School team. Local club team. Travel in the summer.
On “Throughput: The Kids Are All Right but the Surgeon General Isn’t Edition”
ThTh3: I assume that the generative forms of large neural net models will be useful for things eventually. I'm much more interested in the recognition sorts of things that are already useful. For example, significant improvements have been made in optical character recognition accuracy. And stuff that can be built on top of high-quality recognizers.
On “Weekend Plans Post: GenX is, apparently, the last generation to use a top sheet”
I have always been a very warm sleeper. In the summer, most times even a top sheet is too much, I shove all the covers aside and sleep without any of them. In the winter, a top sheet and single blanket is usually enough. Of course, if I'm sharing the bed in the winter -- spouse, pet, whatever -- at some point I wake up because I want to roll over but can't, there's someone snuggled right up against me to stay warm.
On “Five New Rulings from The Supreme Court of the United States”
IIRC, Congress passed specific legislation some time back stating that members of the armed forces may bring their non-citizen spouses into the country.
"
Not for years. Next step is probably to appoint a new special master, since the current one didn't get the states to come to a proper decision. This will be either the fourth or fifth SM for the case, I believe.
From memory, so suspect, but the US objection is that the consent decree didn't settle the issue of whether withdrawals from an aquifer that is hydrologically linked to a surface water flow is a withdrawal from the surface water. The engineering answer to the question has always been yes, with a multiplier between 0 and 1. Most western state water law has always assumed the answer is no. If the Supreme Court ever holds that the engineering answer is the correct one, all hell will break loose.
"
Twenty extended families in the Imperial Valley get more water from the Colorado River than Nevada's entire allocation. The families do a lot of vegetables, but their primary cash crop is alfalfa.
On “Weekend Plans Post: GenX is, apparently, the last generation to use a top sheet”
My father came out of his stint in the US Navy a senior petty officer. Nothing was ever quite clean enough for him, nor we kids' beds made properly. When my wife-to-be and I moved in together, I carefully refrained from remaking the bed after she had done it.
On “Ordinary World: Juneteenth Edition”
Hence the common start-up practice of paying crap, with limited benefits, jammed into minimum space, but with a pile of shares and/or options. Intel millionaires. Microsoft millionaires. Lambda School was almost certainly spending the money differently than on generous salary/benefits for <200 employees.
Nvidia's book value is somewhere over $3 trillion right now. How many of the early people are cashing in and leaving?
"
HOW IN THE HELL DO YOU BURN THROUGH $120 MILLION DOLLARS?
Many years ago -- early 1990s -- I was working at a giant telecom company. The company created a project to build a new billing system without some of the serious limitations of the old system. The project burned through $150 million dollars and produced not a single line of usable code. The company reorganized the technology part of the business for the sole purpose of putting a particular SVP and a particular 200 people into a single organization, fired them all, then three months later reorganized back into something sane.
If you're looking for specific methods, consultants are good. As I recall, the group of consultants that did an incredibly obtuse object-oriented design got a million dollars. When a couple of us who were far enough away from the project that we could safely ask questions asked how big the server farm would be to use the architecture to handle 14 million customers, they said they didn't know but for two million dollars and another nine months they could tell us. Buying hardware because the schedule says it's time to buy the hardware even though there's no actual software architecture is very good.
ETA: Misthreaded. This should be attached to JB's 5:43 pm comment.
"
JT8: I listened to this when it first went up, and the question I was left with was "How many people does it take to go from being an outpost to a colony?" Bases in Antarctica and the ISS are clearly outposts. Basically, they make no effort to become self-sufficient or to rise to the level of technology required to put them in place. Lots of people talk about colonizing the Moon; Elon Musk talks about colonizing Mars; how many people?
The question gets asked, indirectly at least, in a variety of speculative fiction. Heinlein's lunar colonists in The Moon is a Harsh Mistress clearly number in the millions. They're still not self-sufficient. Two examples: manned spacecraft and big computers are come from Earth, not local producers. James Blish's Cities in Flight books imply millions in a couple of different ways. Small planetary colonies drop back to lower technology levels, creating markets for the migrant cities' technology. Even hundreds of cities with populations totaling millions struggle. Exchange of specialists and technology between the cities is a regular occurrence.
My own guess is that the minimum population required to maintain our current level of technology is around 30-40 million. Very few of those people will be involved with the leading edge of technology. Far more of them will be farmers, teachers, nurses, bartenders, etc.
On “Supreme Court Strikes Down “Bump Stock Ban” 6-3”
Based purely on muscle motions, I'd speculate that they pulled the trigger once. Then they held the finger as still and rigid as they could while the trigger banged into it 19 times.
ETA: This is why gun safety courses teach you not to put your finger inside the trigger guard until you are ready to shoot. Because if you just leave the finger there, sometimes the trigger bangs into it and the weapon fires.
On “Open Mic for the week of 6/10/2024”
The wording in the forecast discussion from my local NWS office is almost always a little amusing. Altitude of the area they cover ranges from 4,500 ft to 14,000 ft. In some events, if the storm track is wrong by 50 miles precipitation goes from 4" of snow to 24". A very small error in the altitude of the moisture-bearing layer is the difference between virga and that 0.01" of rain.
When I worked on the U of Colorado campus in Boulder, one of the (probably) academic myth stories that got told was about an intro to meteorology class where students had to turn in their forecast for the next day each time the class met. Some wag supposedly turned in the NWS forecast, and finished next-to-last in the class.
I have read lately that some of the deep machine learning people are building billion-coefficient models for doing forecasts in difficult areas. I would not be surprised if those are more accurate for my local area than the physics-based models.
"
I'm backing the low-probability outcome on this one -- they schedule a rehearing for next term, and put the decision off until after the November election results are known.
"
That's a reasonable thought. Also, given the reaction to Dobbs, five months before the presidential election is not a good time to do a nationwide ban on the most common method of abortion if you want to see more Republicans elected.
On “Time Enough to Last? Age Limiting Members of Congress Passes Ballot in North Dakota”
That would be, at best, term limits in another form. I worked for the Colorado state legislature about the time the state's term limits were starting to really bite. Every two years, after elections, 25-30% of the members were serving their first term. Staff ran training sessions. I was on the budget staff and it was clear from the questions that the new members had absolutely no idea about how the state budget process worked, nor all the constraints on it. It was clear that when the budget staff director was making assignments that he took preservation of institutional memory as a concern, since that memory was no longer going to reside with the members.
On “Open Mic for the week of 6/10/2024”
Being a father isn't something you get to give up when your children reach a specific age. Nor, based on my own experience, is being someone's child something you get to give up. When I had to put my wife in memory care a bit over a year ago, my children's reactions, both in their own way, were "Shut up, Dad, and let us help."
"
I think 538's polling model is missing an important point. Or perhaps the pollsters' models are missing it. From 1988, the Democratic share of the presidential vote in CA, CO, AZ, and NV has increased by between one and two percentage points per four-year cycle. R-squared values from 0.6 to 0.9. I expect Arizona and Nevada will go for Biden where 538's model favors Trump narrowly in those two.
On “Weekend Plans Post: Settling into the Summer”
I'm starting a new applied math project and am in the process of skimming technical journal papers, and reading a subset in detail. The non-political part is "Old mathematicians don't retire, they just go amateur again." The political part, about which I should do a separate post, is that an amateur mathematician on a retiree's budget has to steal the papers via the Russians.
"
AAA has for years offered a buying service, at least in some states. A number of companies advertise that you pick the car and the listed price from their web site and they'll deliver it. Some will also make the deal for your trade-in, in which case the new car comes on a flatbed and they take away the old one.
"
Too close to politics.
"
I had been thinking about replacing my Honda Fit before we had to put my wife in memory care. That costs as much as a modest new car per year, even net of long-term care insurance, so I've reconsidered. Something electric. But no more than I drive, I probably can't even justify it on carbon footprint. New car carbon cost, plus local power won't be carbon-free for several more years vs. my modest (by US standards) gasoline usage. The Fit will be 16 in August. I make it a 50/50 bet whether I have to give up driving before it dies.
On “Safe Nerdy and the Early Adopter Problem”
I did not, and do not now in syndication, watch Big Bang Theory. Summarize it as "These are my people." And I don't care to watch them made fun of.
Of course, I'm the person who was asked "Who are your people, Mike?" at a party where I wasn't paying much attention to the conversation, and answered "The applied mathematicians."
*Comment archive for non-registered commenters assembled by email address as provided.