The problem of shitty software is you because you pay for this shit
prog.nocorp.me
prog.nocorp.me
We always like to think of consumer choice as being the ultimate determiner for what becomes and stays successful, and that's how it should work. But, it's not actually how it works. With sooo many monopolies out there (at&t/comcast ISP), there's a lot of things where you don't have much consumer choice or no consumer. And when you do have a choice, the cost of switching is super high.
I periodically go through my apps and see which ones are taking up the most disk space/memory/bandwidth/battery life and prune all of the ones above a certain amount.
Guess what all the top 10 have in common: yup, they are all pseudo monopolies. If you think the problem is you, I dare you to go to your top 10 biggest wasters and see if you can find a viable alternative to any of them.
Heck, sometimes, it's literally impossible to delete certain apps: the bloatware that's baked into all our phones. But, people don't seem to mind, some of those app even rate 4 stars out of 5.
And it doesn't help that searching for an app is such a big pain point. When you do finally find an app that meets your criteria, it's tough to find another ones that meets All your requirements.
But, my impression is that, most non-tech people don't care about hings like: disk space, memory, bandwidth or effect on battery life. At least, not enough to measure it or do anything about it. My wife would rather carry along a super heavy dumbbell-weight battery, than look at which apps are draining the most juice.
"Your feature made things slow down - you need to fix it!! Oh, BTW we need this next new feature as well".
In markets where individuals do make the software choice, buggy and slow products are often punished. You see a lot of warnings about bugs and performance problems on video game reviews, and those reviews do seem to hurt the sales.
The top game on Steam for many months now is PlayerUnknown's Battlegrounds. It has a "mixed" rating, with only 49% of reviews giving it a thumbs up -- by far the lowest rating of any game in the top 20. The reviews tell a story of how bugs, lag, broken servers, loot crates, and rampant cheating add up to a barely-playable game.
And yet, no one cares! It continues to top the charts, both in sales and cumulative play time. If we think about why that is, we might come up with non-monopoly explanations for why buggy software proliferates in other fields.
It can top charts and have also tuned away a sizeable portion of the potential player base. Both can be true at the same exact time.
You're telling an easy to believe just-so story. And that's the problem with the way these things are analyzed.
But I'm beginning to think I might be wrong, things just continue to get more bloated, and people continue to use it. Some of it is the price. A skype update that now burns a whole CPU core and 400MB ram is still acceptable because its _FREE_.
Long term I suspect I might be right, companies that can run their chat app on 30 servers, vs 3000 will have a competitive edge. Whether I will be around in the decades this will take to shake out is another question.
That alone is an indicator that cost (at least of compute) isn't a critical factor in the decision process. If it were, AWS costing a multiple of self-hosting would render it no longer paradigmatic.
This is probably true for people who only need tiny instances or have massively variable workloads. OTOH, the cloud vendors have been very good at selling people on the idea that its hard to rack machines and buy clean power & network connections. So, even at small companies I've seen AWS bills that could easily pay for another engineer (at middle America prices) despite the fact that a back of the envelope calculation wouldn't lead you to understand where all the compute power is actually going.
Not at all. Although I didn't state it outright (in this thread), I'm convinced that this, rather than cost, is the critical factor in this decision, not just in middle management but all the way to the top (especially in startups where there may not be a middle).
> People regularly talk about how inexpensive AWS is vs buying hardware and renting space.
As I mentioned in another thread [1], they're both right and wrong. For the general case, and if cost were actually a critical factor, they'd be wrong.
However, as you point out, they've been sold on the idea (i.e. perception). Since what they end up having to pay is actual, not perceived cost, perhaps what we're saying is that a form of delusion is a factor in the decision making.
> This is probably true for people who only need tiny instances or have massively variable workloads.
I'd go so far as to say it's definitely true in those cases, but that they're rare/niche enough to be irrelevant to the broader discussion.
Even the latter case, the sweet spot for "elastic" compute, is hard to imagine being the exclusive workload in an environment. To borrow a term from electrical generation, there's always some "base load". If the variable load is being used as justification for running the base load on AWS, then any cost savings could be wiped out [2]. I've found these lines of reasoning never have any actual numbers (cost analysis) behind them.
> the idea that its hard to rack machines and buy clean power & network connections.
Those things are trivial, of course, since they're generally outsourced (and safe [1] to outsource), but there's more to it.
In that way, the idea is also both right and wrong. It's not trivial/obvious how to do it right, without getting bilked for too much money or creating needless risk/instability. It isn't however, cognitively or intellectually challenging and can easily be taught.
It is, however, potentially tedious and unsexy, which means that hiring someone to do it exclusively isn't a credible option. Fortunately, that aspect of even a few hundred servers might take up a quarter of senior engineer's time [3], and all the other things (that would still need doing even with AWS) would take up the rest of the time.
> a back of the envelope calculation wouldn't lead you to understand where all the compute power is actually going.
That's part of the magic of AWS, it doesn't have to "go" anywhere to get charged. GCP recently announced a more pro-rated approach to, at least, CPU cores, based on actual usage.
In general, even to take advantage of elasticity, a customer has to expend (costly) engineering effort and/or engage in the vendor-lock-in auto-scaling tools provided.
Even then, there's opportunities to slip under the radar, such as with unattached EBS volumes, or snapshots on volumes, or S3. I'm sure there's a myriad of other opportunities to get charged for unused services spread throughout the AWS ecosystem.
[1] https://news.ycombinator.com/item?id=18043544
[2] I interviewed with a startup that was attempting to use this justification to move completely out of a colo, even though the variability of their "spiky" load was below 10:1 and their base load was well over half.
[3] Something that is often grossly over-estimated by people who have inadequate experience. I've read here on HN "but then we'd have to hire 5 sysadmins", a 10x over-estimation.
Whatever handful of coins I save going with crappy cheap or free software just isn’t worth the frustration and lost time that’s sure to come as part of the deal.
the problem with open source is clustering of medium-level devs who drive out better devs.
the problem with closed source is management who prefer code monkeys over engineers who want to solve problems.
you as user have zero influence over these phenomenons
>People are more likely to use a buggy app that has features they want over a perfect app with limited features.
I understand the article is specifically addressing the slowness of chat apps..if we’re talking about messenger: I’m sure they could use a fraction of the RAM with the same features. But the fact of the matter is that it has the features people want and we will continue to use it. Is Facebook so in the wrong if their RAM hungry App is so successful? I’m not sure.
On a side note regarding good software, I enjoyed using a site called teamtreehouse so much that I pay for it still 3 years later and I never log in.
When Amazon had the one-click patent, I boycotted them for the better part of a decade. They didn't notice, and I missed out on some of their best service years, when deliveries arrived on time and products were legit
When blizzard cracked down excessively on bnetd, I boycotted them for several years. They didn't notice, and while gaming is not a big part of my life I was a warcraft and starcraft fan that missed out on the original sc1 competitive scene (as a fan, not participant), warcraft 3, and this little thing that grew into DOTA.
Had I treated comcast the same way for their many and continued terrible behaviors, i doubt I'd have the job and experienced I have today.
The issue is not that _I_ pay for this, it is that WE do, and long term vision is a noted human weakness that _I_ lack the power to change no matter how many times I inform the internet that it is wrong (j/k).
I dont have a real solution to offer, but if we think the system is broken, just trying the same ineffective solution over and over won't change it. And if you think the system is great (as the article does)..well, I'll have to disagree. Mass market capitalism has done great things and I wouldn't want to toss it wholesale, but one need only look at the environment (and consider the software environment as well, to keep it topical) to realize there is room for improvement - improvement that doesn't magically happen on it's own when people simply embrace what we have and stop there.
I can't believe people would buy a game where it's literally pay to win. Kids these days...
Seen this a bit, why not create literally an alternate client for Slack if customers actually care? I'm not sure this is a good argument. A large part of Slack is the backend and admin tools and advertising, without these to care about you can catch up fast.
There's the obvious risk being at the mercy of Slack, but any software has risk, getting big corp to pay for your 'better' client is high reward if there actually is value to clients, big corp will pay for it.
Before that it was the free upgrade from win7/8 which was unprecedented. They really want you to be on win 10 to push their store.
Tell that to the police when then check your PC for pirated software...
That makes Debian-flavored Linux my most expensive operating system. I spent a non-trivial amount of time figuring out how that thing works at all!
Actually, I spent way more time on Windows back in the day, so I guess Debian was a relative bargain.