Dan Hosey
@danhosey.bsky.social
HS Physics Teacher. Desmos simulation fan. Spreadsheets too. Always tinkering.
www.mrhosey.com/desmos
www.mrhosey.com/desmos
I couldn't figure out how to quote steve stonebraker's post about this from a couple weeks ago.
October 31, 2025 at 10:47 AM
I couldn't figure out how to quote steve stonebraker's post about this from a couple weeks ago.
I don't use AI for anything I write, not just rec letters. Some people do, but baking it right into the platform tied to detailed academic and demographic data, just feels so wrong to me.
October 31, 2025 at 10:44 AM
I don't use AI for anything I write, not just rec letters. Some people do, but baking it right into the platform tied to detailed academic and demographic data, just feels so wrong to me.
I don't know if this model is trained on the backs of letter writers, but all sorts of models are trained on all sorts of unpaid, uncredited, and often unaware sources. I despise how grey and obfuscated the world has become in the last few years, and how ambiguously commodified we have all become.
October 31, 2025 at 10:40 AM
I don't know if this model is trained on the backs of letter writers, but all sorts of models are trained on all sorts of unpaid, uncredited, and often unaware sources. I despise how grey and obfuscated the world has become in the last few years, and how ambiguously commodified we have all become.
This year I started putting this in my opening paragraph, "In this AI world, I as a human will try to tell you some real things you don't already know about student."
October 31, 2025 at 10:35 AM
This year I started putting this in my opening paragraph, "In this AI world, I as a human will try to tell you some real things you don't already know about student."
I think this AI wave makes the personal anecdotes and specifics,no matter how trivial,stand out.Typos and all.They are the few scraps of humanity left in the process.Also wonder is the model used trained on the millons of letters uploaded to the platform over the years?Is that in a terms of service?
October 31, 2025 at 10:31 AM
I think this AI wave makes the personal anecdotes and specifics,no matter how trivial,stand out.Typos and all.They are the few scraps of humanity left in the process.Also wonder is the model used trained on the millons of letters uploaded to the platform over the years?Is that in a terms of service?
Bonus kudos if you make a car with an attachable sail to increase drag. More kudos if you design the cars to be robust enough to tolerate head on collisions.
October 30, 2025 at 4:23 PM
Bonus kudos if you make a car with an attachable sail to increase drag. More kudos if you design the cars to be robust enough to tolerate head on collisions.
How cool would a hotwheels car with an accelerometer, wheel rotation/linear position/velocity/accerlation,Normal Force sensor be?Kind of like a scaled down smart cart + Normal Force sensor. Imagine making plots of position, speed, centripetal and tangential acceleration, Normal Force around a loop.
October 30, 2025 at 4:21 PM
How cool would a hotwheels car with an accelerometer, wheel rotation/linear position/velocity/accerlation,Normal Force sensor be?Kind of like a scaled down smart cart + Normal Force sensor. Imagine making plots of position, speed, centripetal and tangential acceleration, Normal Force around a loop.
I'm not choked up by that one either.
October 28, 2025 at 6:59 PM
I'm not choked up by that one either.
That's fantastic! Thank you!
October 27, 2025 at 2:31 PM
That's fantastic! Thank you!
I don't know if this model is trained on the backs of letter writers, but all sorts of models are trained on all sorts of unpaid, uncredited, and often unaware sources. I despise how grey and obfuscated the world has become in the last few years, and how ambiguously commodified we have all become.
October 25, 2025 at 3:36 PM
I don't know if this model is trained on the backs of letter writers, but all sorts of models are trained on all sorts of unpaid, uncredited, and often unaware sources. I despise how grey and obfuscated the world has become in the last few years, and how ambiguously commodified we have all become.
I think this AI wave makes the personal anecdotes and specifics,no matter how trivial,stand out.Typos and all.They are the few scraps of humanity left in the process.Also wonder is the model used trained on the millons of letters uploaded to the platform over the years?Is that in a terms of service?
October 25, 2025 at 2:29 PM
I think this AI wave makes the personal anecdotes and specifics,no matter how trivial,stand out.Typos and all.They are the few scraps of humanity left in the process.Also wonder is the model used trained on the millons of letters uploaded to the platform over the years?Is that in a terms of service?
I saw your post, looked and sure enough, there it is when you go to upload a letter. I agree. I don't use AI for anything I write, not just rec letters. Some people do, but baking it right into the platform tied to detailed academic and demographic data, just feels so wrong to me.
October 25, 2025 at 2:29 PM
I saw your post, looked and sure enough, there it is when you go to upload a letter. I agree. I don't use AI for anything I write, not just rec letters. Some people do, but baking it right into the platform tied to detailed academic and demographic data, just feels so wrong to me.
I'd love a native way to control the bounds of the axes to variables, similar to the way we can change the dimensions and orientation of a picture with variables.
October 25, 2025 at 2:16 PM
I'd love a native way to control the bounds of the axes to variables, similar to the way we can change the dimensions and orientation of a picture with variables.
Could also be fun to randomly drop 3 known masses around the center of a bull's eye, students drag the masses to where they think they need to be to get close to the center of the bull's eye. You'd have to restrict the masses to some minumum distance. Then they get a score when their cm is revealed
October 24, 2025 at 3:42 PM
Could also be fun to randomly drop 3 known masses around the center of a bull's eye, students drag the masses to where they think they need to be to get close to the center of the bull's eye. You'd have to restrict the masses to some minumum distance. Then they get a score when their cm is revealed
I just missed the dimension slider the first time I read. I see you already included 2d.
October 24, 2025 at 3:38 PM
I just missed the dimension slider the first time I read. I see you already included 2d.
Could have the user drag the angle of the block and release, then animate it falling one way or the other. Could have the block at an angle fill with water and watch the cm go up until it tips over.
October 24, 2025 at 3:00 PM
Could have the user drag the angle of the block and release, then animate it falling one way or the other. Could have the block at an angle fill with water and watch the cm go up until it tips over.
Could have students predict the cm location by dragging a predicted location, then reveal the actual cm and show them numerically how close they got. A sort of game.
October 24, 2025 at 3:00 PM
Could have students predict the cm location by dragging a predicted location, then reveal the actual cm and show them numerically how close they got. A sort of game.
Very nice. You could have a "not flatten" mode where the y-coordinates also vary. Complete tangent, but it could be fun to made a simulation of a block tipping over. Imagine a block you could change the angle of. Show the CM and weight vector, maybe a vertical line at the corner.
October 24, 2025 at 3:00 PM
Very nice. You could have a "not flatten" mode where the y-coordinates also vary. Complete tangent, but it could be fun to made a simulation of a block tipping over. Imagine a block you could change the angle of. Show the CM and weight vector, maybe a vertical line at the corner.
I think it is common to talk about accumulated error in a numerical simulation. I think of simulations with a time step. For example Runge Kutta drastically reduces the error with each time step in an orbit simulation, or a large angle pendulum. I am curious what sparked your question.
October 21, 2025 at 2:14 AM
I think it is common to talk about accumulated error in a numerical simulation. I think of simulations with a time step. For example Runge Kutta drastically reduces the error with each time step in an orbit simulation, or a large angle pendulum. I am curious what sparked your question.
In calculus based (AP Physics C) Electricity and magnetism, students learn about Gauss's Law,which we use here for radially symmetric charge where charge density varies with radius.I wanted a way to create a visual of charge density in a sphere varying with radius.The sim above matches the expected
October 15, 2025 at 12:06 AM
In calculus based (AP Physics C) Electricity and magnetism, students learn about Gauss's Law,which we use here for radially symmetric charge where charge density varies with radius.I wanted a way to create a visual of charge density in a sphere varying with radius.The sim above matches the expected
For high school physics? incline,pulley,meterstick,stop watch,tiny bowling ball pendulum,ferris wheel,2 cars colliding, 2 vectors and their resultant,tiny mass on a tiny spring, a gyroscope,balancing bird,lens, a diffraction grating,small light bulb and battery,2 magnets,double rainbow,f=ma card
October 13, 2025 at 4:25 PM
For high school physics? incline,pulley,meterstick,stop watch,tiny bowling ball pendulum,ferris wheel,2 cars colliding, 2 vectors and their resultant,tiny mass on a tiny spring, a gyroscope,balancing bird,lens, a diffraction grating,small light bulb and battery,2 magnets,double rainbow,f=ma card