4 ms·

> we accept bloat as a natural fact of life and accommodate to that I don’t disagree with your post, but what is the solution here? Your average consumer like

> we accept bloat as a natural fact of life and accommodate to that I don’t disagree with your post, but what is the solution here? Your average consumer likely doesn’t understand the specific role of computer RAM, nor do they understand that software they’re using excessively wastes it. How can they accept or reject something they don’t know? And it’s a similar problem with many developers. They don’t think about memory, or they view it as some infinite resource pool. I see folks often times using complex “frameworks” to solve trivial problems. They merely wanted to 2 + 2, but that framework has a dependency graph for solving systems of equations and partial derivatives. Democratization of tech and programming has resulted in a lot of inefficiencies. I mean - that’s the cost, right? > we encourage active obscolecence and further pollution of our environment I don’t think your typical consumer encourages this at all. Between their jobs, families, and all the stress of day to day life, Im not convinced they even have time to think about these things.

alephnan 5y ago
Apple’s wall garden somewhat curates a high quality set of apps
- MomoXenosaga 5y ago
  Sure but those apps are often also ridiculously restrictive compared to their Android counterparts.
  - rql10 5y ago
    In that they restrict you from screwing around as root and otherwise aren’t “ridiculously restrictive” compared to Android in any way? You're right, they are.
jcelerier 5y ago
> but what is the solution here ? Like always for consumer protection, the only recourse is legal action. Ask your govt representatives to make slow software illegal. Carbon tax on inefficient code. IIRC Germany has started looking into it ?
- rql10 5y ago
  “We can’t fix software, so let’s call the police.”
  - jcelerier 5y ago
    It is not a technological issue, but a social one - the incentives to create consumer-friendly software are not good enough so companies settles on the lowest possible denominator.
  - rakoo 5y ago
    It's not calling the police, it's changing the law to something people agree. It's the basis of any society
- Jiocus 5y ago
  I've read about manufacturers limiting RAM to e.g. 8GB to be able to stay within some power-consumption deemed acceptable for a notebook device. More RAM draw power. Given that RAM size can have those ramifications energy-wise, it sounds almost like it would be a win, to go against the grain of the article and push for an upper limit of memory, say 4-8GB, if the environmental factor was the lone metric to judge on. Secondary effects follows due to restraints, software makers who doesn't want to see users churn from their applications start scrambling to optimize codebases so the executives won't loose out on their bonuses for that fiscal year.
- tracer4201 5y ago
  I think this could be a can of worms that’s more damaging. What if the software is completely usable when it’s slow? What if it’s slow by design? That’s not outlawed? And what’s the metric for being slow? Latency? Latency across what dimension? Who measures it and policies this exactly? Or do we slap fines based on resource consumption? What if my software just runs slower on the JVM if I’m on Java 11 instead of Java 8? What if that problem isn’t my fault but because someone in my dependency graph has a regression? Oh they open sourced their code. Am I liable now because I have a dependency on it and upgraded to the new version? What if it’s network conditions that are causing the slow performance. Am I liable? Is your internet provider liable? What’s the minimum bar for performance, across what dimensions, and who defines it? And where do we find people with the education, background, training to define such a vast standard? How do we ensure it’s fair? Maybe I can’t write as performant code as the well funded people at Google or Amazon. Am I just shut out of the market entirely? Or is it applied selectively based on market cap of a company?
toast0 5y ago
> I don’t disagree with your post, but what is the solution here? If you provide hardware for developers who build software with a direct user interface, don't get them the best. Getting them the best for build machines, so they don't have to wait around for builds makes sense, but make sure the machines they run interactive tests and personal usage on are not the best. When your developers have 16gb or more ram with 8 cores and a fast SSD and your users have 4 gb, a dual core 15w cpu, and a spinning disk, it can be hard to understand the sometimes huge difference in experience. And sometimes, it's actually not that much work to make things better for users with low end devices. The difference between high end phones and low end phones can be even more stark.
- seanp2k2 5y ago
  On my last team, we would do QA on devices that ranged from “decade old + notoriously buggy implementation” to “new hotness, not even available to consumers yet”. The latest greatest stuff was more useful so the mfgs could understand what wasn’t done right there, but the old stuff is where our SDK team found the most value — if it ran fine there, it was pretty much guaranteed to be awesome everywhere else. Faster devices would just mask things like frame drop / memory leaks, but those good old low-end old devices were great for detecting problems and perf regressions. An anecdote that has long stuck with me from a bicycle mechanic back in the rust belt: ~“Anyone can make new stuff work well, but not many folks still have the patience and expertise to really make the older stuff shine, but it can and it does with a bit of care.” For work, I upgraded from an 8th Gen Intel 6-Core i9 laptop with 32GB@2400mhz and a 1050ti to a 5950x / RTX 3090 / 32GB@3600mhz with tight timings and it’s very noticeably faster for general desktop usage and development.
  - rakoo 5y ago
    Facebook implemented "2G Tuesdays" (https://www.nbcnews.com/tech/social-media/facebooks-2g-tuesdays-slow-down-internet-employees-n452466 https://www.nbcnews.com/tech/social-media/facebooks-2g-tuesd...) where developers would get 2G speeds when doing anything related to the website. That gives them the experience of people with very poor connectivity and probably improve their experience. Just like you said, if it runs OK on 2G, it can only be fast enough on 4G

alephnan 5y ago

Apple’s wall garden somewhat curates a high quality set of apps

MomoXenosaga 5y ago
Sure but those apps are often also ridiculously restrictive compared to their Android counterparts.
- rql10 5y ago
  In that they restrict you from screwing around as root and otherwise aren’t “ridiculously restrictive” compared to Android in any way? You're right, they are.

jcelerier 5y ago

> but what is the solution here ? Like always for consumer protection, the only recourse is legal action. Ask your govt representatives to make slow software illegal. Carbon tax on inefficient code. IIRC Germany has started looking into it ?

rql10 5y ago
“We can’t fix software, so let’s call the police.”
- jcelerier 5y ago
  It is not a technological issue, but a social one - the incentives to create consumer-friendly software are not good enough so companies settles on the lowest possible denominator.
- rakoo 5y ago
  It's not calling the police, it's changing the law to something people agree. It's the basis of any society
Jiocus 5y ago
I've read about manufacturers limiting RAM to e.g. 8GB to be able to stay within some power-consumption deemed acceptable for a notebook device. More RAM draw power. Given that RAM size can have those ramifications energy-wise, it sounds almost like it would be a win, to go against the grain of the article and push for an upper limit of memory, say 4-8GB, if the environmental factor was the lone metric to judge on. Secondary effects follows due to restraints, software makers who doesn't want to see users churn from their applications start scrambling to optimize codebases so the executives won't loose out on their bonuses for that fiscal year.
tracer4201 5y ago
I think this could be a can of worms that’s more damaging. What if the software is completely usable when it’s slow? What if it’s slow by design? That’s not outlawed? And what’s the metric for being slow? Latency? Latency across what dimension? Who measures it and policies this exactly? Or do we slap fines based on resource consumption? What if my software just runs slower on the JVM if I’m on Java 11 instead of Java 8? What if that problem isn’t my fault but because someone in my dependency graph has a regression? Oh they open sourced their code. Am I liable now because I have a dependency on it and upgraded to the new version? What if it’s network conditions that are causing the slow performance. Am I liable? Is your internet provider liable? What’s the minimum bar for performance, across what dimensions, and who defines it? And where do we find people with the education, background, training to define such a vast standard? How do we ensure it’s fair? Maybe I can’t write as performant code as the well funded people at Google or Amazon. Am I just shut out of the market entirely? Or is it applied selectively based on market cap of a company?

toast0 5y ago

> I don’t disagree with your post, but what is the solution here? If you provide hardware for developers who build software with a direct user interface, don't get them the best. Getting them the best for build machines, so they don't have to wait around for builds makes sense, but make sure the machines they run interactive tests and personal usage on are not the best. When your developers have 16gb or more ram with 8 cores and a fast SSD and your users have 4 gb, a dual core 15w cpu, and a spinning disk, it can be hard to understand the sometimes huge difference in experience. And sometimes, it's actually not that much work to make things better for users with low end devices. The difference between high end phones and low end phones can be even more stark.

seanp2k2 5y ago
On my last team, we would do QA on devices that ranged from “decade old + notoriously buggy implementation” to “new hotness, not even available to consumers yet”. The latest greatest stuff was more useful so the mfgs could understand what wasn’t done right there, but the old stuff is where our SDK team found the most value — if it ran fine there, it was pretty much guaranteed to be awesome everywhere else. Faster devices would just mask things like frame drop / memory leaks, but those good old low-end old devices were great for detecting problems and perf regressions. An anecdote that has long stuck with me from a bicycle mechanic back in the rust belt: ~“Anyone can make new stuff work well, but not many folks still have the patience and expertise to really make the older stuff shine, but it can and it does with a bit of care.” For work, I upgraded from an 8th Gen Intel 6-Core i9 laptop with 32GB@2400mhz and a 1050ti to a 5950x / RTX 3090 / 32GB@3600mhz with tight timings and it’s very noticeably faster for general desktop usage and development.
- rakoo 5y ago
  Facebook implemented "2G Tuesdays" (https://www.nbcnews.com/tech/social-media/facebooks-2g-tuesdays-slow-down-internet-employees-n452466 https://www.nbcnews.com/tech/social-media/facebooks-2g-tuesd...) where developers would get 2G speeds when doing anything related to the website. That gives them the experience of people with very poor connectivity and probably improve their experience. Just like you said, if it runs OK on 2G, it can only be fast enough on 4G