SafariMonkey
@safarimonkey.bsky.social
Working on open source XR at @collabora.com. Rust enthusiast. pfp @jfrxd on Twitter. they/he
The "marginal cost of intelligence" is an important question, especially if you consider current models interesting demos but insufficiently consistent to be left unsupervised enough to be cost-effective.
October 10, 2025 at 6:33 PM
The "marginal cost of intelligence" is an important question, especially if you consider current models interesting demos but insufficiently consistent to be left unsupervised enough to be cost-effective.
Fair. I think mentioning this in the post would have aided your argument, because your point appeared to me to be, "as long as the newer models are a bit better, it doesn't matter if they're much more expensive". I came into the post on your side and was surprised that that was your best argument.
October 10, 2025 at 6:29 PM
Fair. I think mentioning this in the post would have aided your argument, because your point appeared to me to be, "as long as the newer models are a bit better, it doesn't matter if they're much more expensive". I came into the post on your side and was surprised that that was your best argument.
When past models were sold at least partly on the promise of, "switch your workflow now, it'll only get better", I'd say it's relevant when "better" is substantially more expensive.
Also, colloquially, people do say e.g. GPUs are getting more expensive, even though you get more perf per dollar now.
Also, colloquially, people do say e.g. GPUs are getting more expensive, even though you get more perf per dollar now.
October 10, 2025 at 4:41 PM
When past models were sold at least partly on the promise of, "switch your workflow now, it'll only get better", I'd say it's relevant when "better" is substantially more expensive.
Also, colloquially, people do say e.g. GPUs are getting more expensive, even though you get more perf per dollar now.
Also, colloquially, people do say e.g. GPUs are getting more expensive, even though you get more perf per dollar now.
still happening for me...
October 5, 2025 at 8:44 PM
still happening for me...
so one thing I did figure out is that restarting, unblocking add-ons, restarting, and then launching worked for me... I've heard the add-ons can be done by unsetting a flag in steamvr.vrsettings, avoiding a restart, but I haven't tried that yet.
October 5, 2025 at 5:29 PM
so one thing I did figure out is that restarting, unblocking add-ons, restarting, and then launching worked for me... I've heard the add-ons can be done by unsetting a flag in steamvr.vrsettings, avoiding a restart, but I haven't tried that yet.
I just wish it also knew that I'd already seen it. I frequently see such a video with no progress marking and I know I've seen it, and/or open it and find that I've already liked some of the comments.
September 20, 2025 at 2:18 AM
I just wish it also knew that I'd already seen it. I frequently see such a video with no progress marking and I know I've seen it, and/or open it and find that I've already liked some of the comments.
I guess unless they're disabled, or unless the VRAM is used by things other than textures (do e.g. meshes with an insane number of blendshapes count towards the calculated VRAM number?)
August 29, 2025 at 2:50 PM
I guess unless they're disabled, or unless the VRAM is used by things other than textures (do e.g. meshes with an insane number of blendshapes count towards the calculated VRAM number?)
"New web browser or Chromium again?"
"...Chromium again."
"...Chromium again."
July 10, 2025 at 5:29 PM
"New web browser or Chromium again?"
"...Chromium again."
"...Chromium again."
stray thought: your perspective will change, but that's because you haven't had all the experiences you will have. at each step, you do your best to make choices based on the experiences you have so far, and you learn. as for trust, sometimes you're the only person you can trust to look out for you.
June 27, 2025 at 2:51 AM
stray thought: your perspective will change, but that's because you haven't had all the experiences you will have. at each step, you do your best to make choices based on the experiences you have so far, and you learn. as for trust, sometimes you're the only person you can trust to look out for you.
I still remember when FireBox was renamed to JanusVR. Really cool project, loved the non-centralised hosting, and being able to just walk through portals made it really feel like an interconnected space.
June 25, 2025 at 9:20 AM
I still remember when FireBox was renamed to JanusVR. Really cool project, loved the non-centralised hosting, and being able to just walk through portals made it really feel like an interconnected space.
indeed! yes, a logarithmic scale is very reasonable, I'm sorry that I didn't clarify that.
June 24, 2025 at 3:49 PM
indeed! yes, a logarithmic scale is very reasonable, I'm sorry that I didn't clarify that.
The problem is on any logarithmic scale, going down by a constant amount will represent dividing the measured quantity by some constant, but you cannot do a finite number of divisions by a finite amount and reach 0. (Scientific notation is almost a log scale, so ignoring the mantissa, 10^what = 0?)
June 24, 2025 at 3:25 PM
The problem is on any logarithmic scale, going down by a constant amount will represent dividing the measured quantity by some constant, but you cannot do a finite number of divisions by a finite amount and reach 0. (Scientific notation is almost a log scale, so ignoring the mantissa, 10^what = 0?)
Exponential and logarithmic functions are inverses of each other. Assuming you mean that incremental additional Kelvin at high temperatures is represented by a smaller change than at low temperatures, I would call that a log scale, and a linear increase in that scale would be exponential in Kelvin.
June 24, 2025 at 3:20 PM
Exponential and logarithmic functions are inverses of each other. Assuming you mean that incremental additional Kelvin at high temperatures is represented by a smaller change than at low temperatures, I would call that a log scale, and a linear increase in that scale would be exponential in Kelvin.
With a logarithmic scale, is it even possible to set a value for absolute 0? I would think it would have to "be" at -inf
June 24, 2025 at 8:43 AM
With a logarithmic scale, is it even possible to set a value for absolute 0? I would think it would have to "be" at -inf
This actually happened to me with a CO detector with low battery on the cupboard right next to the fire alarm. We were trying to figure out what was wrong with that alarm for over a week.
June 11, 2025 at 10:14 AM
This actually happened to me with a CO detector with low battery on the cupboard right next to the fire alarm. We were trying to figure out what was wrong with that alarm for over a week.
One I've heard is decent is Ancients(US)/Relics(EU), groups for adults to make friends, but I've personally never been. Otherwise, I've dropped by "bar" groups that were pretty chill, made a few friends. No idea what would appeal to you, but my point is that publics are not your only option.
June 5, 2025 at 6:41 PM
One I've heard is decent is Ancients(US)/Relics(EU), groups for adults to make friends, but I've personally never been. Otherwise, I've dropped by "bar" groups that were pretty chill, made a few friends. No idea what would appeal to you, but my point is that publics are not your only option.
💯, that's why I mostly find new connections through my existing friend circle. What I would do if you do want to try it is to look for events that appeal to you and find people you get on with there. For example, I found a lot of my first friends through the (now defunct) Transhumanists in VR group.
June 5, 2025 at 6:27 PM
💯, that's why I mostly find new connections through my existing friend circle. What I would do if you do want to try it is to look for events that appeal to you and find people you get on with there. For example, I found a lot of my first friends through the (now defunct) Transhumanists in VR group.
I see publics in VRC as like big public Discord servers. They're loud, but not where most of the real value is. Once you find some friends you vibe with, and then their friends, that's when you (hopefully) find your people.
June 5, 2025 at 6:18 PM
I see publics in VRC as like big public Discord servers. They're loud, but not where most of the real value is. Once you find some friends you vibe with, and then their friends, that's when you (hopefully) find your people.
It has about 2x the graphical performance! And I can't find good CPU benchmarks, but I see claims of +33% single core (but -25% cores, so multicore should be similar).
June 5, 2025 at 6:02 PM
It has about 2x the graphical performance! And I can't find good CPU benchmarks, but I see claims of +33% single core (but -25% cores, so multicore should be similar).
Maybe it's not as important for ancient cultures, but inaccurate/stereotypical portrayal of modern cultures has a real effect when it's many people's sole exposure to them.
June 3, 2025 at 6:23 AM
Maybe it's not as important for ancient cultures, but inaccurate/stereotypical portrayal of modern cultures has a real effect when it's many people's sole exposure to them.
my point is that the reasons for lack of cross-compat between PSVR and PCVR are not technical, they're business decisions.
June 3, 2025 at 12:13 AM
my point is that the reasons for lack of cross-compat between PSVR and PCVR are not technical, they're business decisions.
Console exclusives have existed forever.
June 2, 2025 at 11:41 PM
Console exclusives have existed forever.
I'm surprised that you use -1 instead of +1 for the lower bound. In my mind:
x <..< y
would ideally be read as an expression, meaning the range is all values satisfying
x < n < y
which would make the lowest value of n x+1, and the highest value of n would be y-1. = would mean ≤ instead of <.
x <..< y
would ideally be read as an expression, meaning the range is all values satisfying
x < n < y
which would make the lowest value of n x+1, and the highest value of n would be y-1. = would mean ≤ instead of <.
May 30, 2025 at 3:53 AM
I'm surprised that you use -1 instead of +1 for the lower bound. In my mind:
x <..< y
would ideally be read as an expression, meaning the range is all values satisfying
x < n < y
which would make the lowest value of n x+1, and the highest value of n would be y-1. = would mean ≤ instead of <.
x <..< y
would ideally be read as an expression, meaning the range is all values satisfying
x < n < y
which would make the lowest value of n x+1, and the highest value of n would be y-1. = would mean ≤ instead of <.
I think one reason people are so hostile about it is that the development and adoption of the technology is continuing at full speed almost regardless of their concerns, so the value of any slow and thoughtful discussion seems outweighed by the expansions made by the AI industry in the same time.
May 28, 2025 at 7:43 PM
I think one reason people are so hostile about it is that the development and adoption of the technology is continuing at full speed almost regardless of their concerns, so the value of any slow and thoughtful discussion seems outweighed by the expansions made by the AI industry in the same time.