wtracy

> Sometimes it feels as though existing problems get a free pass because of tradition

In safety-critical systems, known problems are often tolerated because they are predictable. In these cases, mitigations are understood, and there's a well-defined upper bound on the amount of damage that can be caused if the mitigations fail.

> but new issues (even if addressed) are scary and so the entire thing should be stopped.

New issues are unpredictable, and do not have known mitigations. (Once they can be reliably predicted and mitigated, they are no longer "new".) There's also no known upper bound on the amount of damage these issues can cause.

As an added bonus, the quantity of bugs in a long-deployed system is generally well understood, while it's difficult to place an upper bound on the number of bugs in a system that hasn't been tested in production yet.

To make this concrete: Imagine finding a bug that causes several thousand phones to occasionally reboot unexpectedly. Pleased with yourself, you publish a patch and push it out to all affected devices.

One week later, a thousand of those devices power off and never power on again--they've been permanently bricked by your update.

Are your users angry because you refused to give the random reboots a "free pass" because of "tradition"? Or are they angry because you made drastic changes to a system that basically worked without taking the time to understand the consequences?

In this case, I imagine that a "hard fork" would entail starting a new smart contract and copying over the old balances.

Since I'm on mobile and can't test it right now: Do you support export to other platforms/formats? Is there an easy way to get the finished video onto, say, YouTube?

If it's worth anything, my impression is that the CPU (and probably GPU) performance of chromebooks varies widely between models. It seems to be particularly dramatic when comparing ARM models against Intel and AMD models.

I don't doubt that low-end CPUs could cause dropped frames during video recording.

If I had to offer any advice, it would be to test as many models as you can and then decide which ones to officially support.

> Has apple ever blocked or threatened anyone for developing a rival OS?

The iPhone boot ROM will only launch operating systems cryptographically signed by Apple:

https://support.apple.com/guide/security/hardware-security-o...

I'm trying to do something similar with Orgmode right now, and I find it very helpful.

The article touches on something I've been meaning to explore more deeply: Grouping tasks by "triggers". For example, tasks that need to be repeated every week, tasks that need to be done on the next weekday, tasks that have to wait for the pandemic to end first, and so on.

Thinking out loud:

Since handles refer to offer rather than absolute addresses, this sort of scheme should allow for "serializing" data by memmapping chunks of memory to disk.

Obviously, this scheme wouldn't be useful for cross-platform save files, and might not even translate across builds.

Still, as long as you have a robust scheme for invalidating data when a new version is deployed, I could see this being useful for caching, and for saving state for short periods of time on mobile platforms that like to restart processes with little warning. (Rather than potentially losing your place within the game level any time you switch between apps, you only lose your place within the game level when an app update gets installed.)

> wouldn't every website allow everything Legal and be able to just point at the laws and say, "if you don't like it, change the law"?

Again and again, industries operated by rational agents self-regulate with the goal of preventing the government from getting involved. (With many obvious exceptions, like businesses seeking regulatory capture, producers of commodities with razor thin margins, operators who don't expect to be in business long enough for regulation to take effect, etc.)

Once you get the government involved, you have not only the risk of public backlash whenever you make a mistake, you have the risk of fines or worse. Now your legal council has to get involved in the operation of your moderation system. You potentially need multiple people to sign off on moderation decisions to ensure compliance as well as a system to maintain a paper trail in case government auditors come sniffing around. Your operational costs begin to balloon.

A PR person who says, "If you don't like our behavior, change the law" is basically self-destructive.

If that's the reason, it would be nice for Google to come out and say so.

> Because, let's face it, nobody in their right mind procrastinates if their livelihood is on the line or if it means letting someone down.

I see you've never met a person with impaired executive function.

It's frustrating and heartbreaking to watch an otherwise intelligent and capable person just...stop living.

If all you have to go by is the popular stereotypes, you will be shocked by the number of engineers with ADD.

There are third-world sweatshops that solve capchas for a fraction of a penny. I imagine they provide an API, so your bot makes an http call and gets a solution back.

> Bad decision-makers who make slow and thoughtful decisions are still bad decision-makers. They need to make better decisions.

Is this gpt-3 output or something?

Otherwise, if the best insight this guy can come up with is that you can make more good decisions by making better decisions, then I am concerned.

You can silence the alarm by pulling the battery connection. Obviously, that's inconvenient if you want the UPS to be useful again as soon as the power comes back.

I personally like the white noise from electronics, so I can't help you there.

Opening Medium in an incognito window has so far away been enough for me to get past that.

Also, are we due for a snappier phrase than "requires registration"? Loginwall, registerwall, accountwall?

I think you're overstaying things a bit, but you do have a point.

She probably has never experienced poverty, and no doubt believes (probably correctly!) that she could easily walk out and get another job for a similar (or higher) wage.

To build on that: Jobs was absolutely someone who did everything "for the love of the game." It's just that the game that interested him was marketing and design, not engineering.

I can't imagine Jobs as someone motivated put his head down and do busywork to collect a brass nameplate with a fancy title after twenty years.

Several people who dislike the term "software engineer" have started using "software craftsman" instead.

At least one person tried this and found it to be a very hard sell: https://tjcx.me/posts/i-wasted-40k-on-a-fantastic-startup-id...

I think parent was complaining that there's very little public information (particularly hard numbers) about the incumbents. GP was cruising Wikipedia for information about the apps' parent companies rather than just trying to identify the apps.

It's easy to see how many apps are in the store. It's a lot harder to know how much money they are grossing.

Which is why a lot of the real money is in products that have wide appeal for non-disabled users.

This is where most of the bizarre "as seen on TV" products come from: The inventor made a tool for people with disabilities, found that the target market wasn't large enough to sustain a business, and frantically tried to pivot. Pivoting is hard when you already have a large inventory of hard products.

If you can build something that able-bodied people are not only willing to use but willing to pay extra for, then you're good to go.

I'm not so sure about your example.

Get a self-driving truck, and add some sort of people-sensing system that allows potential customers to intuitively flag it down. Then add a self-serve ice cream vending machine.

Now you can build a fleet of ice cream trucks that don't need human operators. I think you could make a few millionaires with that.

My point is not that automated ice cream trucks are the future. My point is that "small" improvements amortized across enough units can be significant.

Now, there's probably a pretty hard ceiling to the amount of income one of these trucks can bring in, so investing in business intelligence or a snazzy customer experience probably isn't a good investment. (Which is probably what parent was imagining.)

But don't underestimate the power of shrinking costs. If you sell a billion screws a year, a one-tenth of a cent improvement in unit manufacturing cost is a million dollars a year.

Who wants to bankroll me if I go create a rural ISP somewhere (probably Nevada/Utah/Arizona, where regulation is lax), become a local hero for providing broadband communication where the incumbents won't, then throttle/censor Fox/Breitbart (maybe requiring a "premium subscription" for access) and tell everyone who complains, "Hey, it's legal!"

Think maybe we could actually get the Republicans on board with net neutrality?

I remember Food 4 Less back in the day (maybe twenty years ago) having a sort of hybrid system where the cashier scanned your items and rang then up, but you bagged your own groceries.

Are you familiar with any similar systems today? How do they rate on throughput?

Understanding a culture is not a binary thing.

Five years is enough time to understand a lot more than someone who has never set for in the country. Naturally, there's other things that they still wouldn't understand after twenty years.

Other companies are failing because they're making mistakes that someone who's been in the US less than a year could recognize. ("Don't use imagery of white and black people facing off to advertise that your product comes in white and black colors.")

They "hit a wall" in the sense that they couldn't keep up the hockey stick style growth that venture capitalists have come to expect.

Just thinking out loud: Wouldn't it be possible to get much more accurate temperature readings by having people open wide and imaging the backs of their throats? This could still be done at a reasonable distance.

Of course, that would require people to stop long enough to take a reading, and would be useless for the products trying to scan crowds with an unattended camera and AI.

And, of course, nothing is really going to help you if the thermometers are inaccurate to begin with.

> I also wonder if you'd even bother writing code to do it on the GPU in the first place.

Some of your customers have large datasets that are only manageable with hardware acceleration, and others don't? Maybe you don't want to maintain two separate codebases?

Being able to run two images per second through a post-processing and/or computer vision pipeline still sounds pretty useful to me.

Shoot, Blender still supports users without accelerated graphics. (Or did five years ago, things may have changed.)

> These guys can scale up by hiring 100s

If you live in the third world, that immediately puts you on the radar of corrupt government officials and organized criminal organizations that want to extort you.

The ransomware gig allows you to stay small and keep a low profile. Nobody has to know that you are raking in the cash until you pack up your family and book it to Tahiti.

> With their current approach they will almost certainly get traced eventually and end up in jail.

Prison in a first world country is a lot better than getting executed, watching your family get executed, or getting kidnapped and forced into slave labor for the local mafia boss.