In an age where we have more access to information than ever, counterfeiting is easier than ever. This makes it difficult to establish and maintain a "reputation" for good product. The European solution this problem is country-of-origin laws for certain categories of product. But even then, if 1000 glass bottles with a "Champagne" label is worth $30,000, then some counterfeiter is going to figure out how to add the labels of a fake winery. So there is an ever increasing labelling race in Europe to try and develop better labels to detect fakes.

The problem is world-wide and affects entire categories of products. This is an article describing how all 16TB SSDs on Amazon are fakes. In some categories, there is so much knock-off product on Amazon, that people aren't sure what is a good deal and what isn't. Thus even if someone has a genuine "great" product, it is tough to differentiate it from complete crap. The only solution for the customer is to assume it is all cheap fakes. This squeezes out the good vendors, increases the proportion of counterfeits, and further worsens the customer experience.

I never really understood why it was in Amazon's interest to sell fake stuff. Amazon must have done a calculation that more money is to be made selling knock-offs than good reputable products.

I remember asking a cd duplicating company about there copyright policy. They informed me that they don't duplicate copyrighted material. When I asked how that worked fir the original author, they seemed quite certain of it.

Any good printing company will be paying attention to what they are printing. They are also likely checking that they aren't unintentionally violating copyright law. So take the privacy claims with a grain of salt ...

It seems like the question fell victim to two things:
1. The question is longer than the answer. English is so imprecise that it becomes difficult to accurately understand the question under pressure. There are so many ways that it is possible to code roughly the same thing, that it is very easy to miss a subtle detail. For this question, there are lots of details.
2. Last question on a time-limited test.

I am less than 2 kms from city hall in a city with almost 1 million people. My fastest "advertised" upload speed is 30 Mb/s. I am a long way from a symmetrical 1GB connection ...

Given the state of things in town, I can't imagine what its like for the truly rural folk.

The number of places that you can get a 1GB internet from two different providers in North America is pretty small. If you can only get it from one provider, then that provider isn't going to try too hard and will gladly extort you for the privilege.

And I mean an actual 1 GB continuous, not 1GB/s for 0.5s ...

The goal function is even more important than the training data. The goal of this AI was to maximize minutes by generating the longest possible conversation. It's not particularly bright, so it can't go with an intelligent conversation. So it is trying to maximize returns by going for the longest unintelligent conversation with a male audience. Sexual content is the result of the optimization.

For what I do, I put most of my effort in the goal function. Otherwise the optimization results in a completely different outcome than what might be expected ...

It's not that hard to get published. The hard part is delivering it at a conference with a straight face.

In the humor section.

There are Energizer, Duracell, Republicans and Democrats. All agree they need campaign donations to keep DST.

The slogan "change the batteries in your smoke detector every time the time changes" sells so many batteries that the battery companies make sure the DST bill does not pass congress.

New building code specifies line powered smoke detectors, and many newer smoke detectors come with batteries good for 10 years. So it is possible that technology will save us ...

The issue is that if selling a sofa for $600 qualifies as business income then it also has business expenses associated with it, and it is likely a money losing transaction. The resulting administrative and accounting expenses, fights with the IRS, audits, and legal expenses are likely to cost more than the resulting revenue.

It will be a clear win for the lawyers and accountants involved. The additional paperwork should generate significant revenue, which can definitely be claimed as a business expense.

I can't escape feeling that NASA asked Space/X to wait, because they don't want SLS upstaged too badly.

As it is, everyone seems to think Starship is the future, even though it hasn't flown yet.

This could come back on Intel. "Dear Intel, we would like to discuss this patent suit. We purchase $x in Intel PCs per year. ..."

AMD and Intel are cross-licensed to the hilt. They can't do anything without each others patents. I believe most of the other major chip players (Nvidia, ARM, TSMC) are too. They should be safe.

QuickC was fast, and it highlighted the line on which an error occurred.

Today, we have three choices:
a) Use VisualStudio / Visual C++, which will be slow, but will highlight the line on which an error occurs.
b) Use VIM, which is fast. It is fast to open, and quick to type ":1000" to get to the relevant line number.
c) Use a language like Python which associates an error with a multiple line numbers that sometimes have nothing to do with the line that needs changing.

I have often wondered if we could add the following features to VIM:
a) Code completion
b) The error/file number feature.
c) A somewhat intelligent background compile/syntax check?
without slowing the entire program down. That would give me 90% of the features of a full blown IDE, with all of the speed of a lightweight editor like VIM.

I'm not even that fussy about VIM, and routinely switch back and forth between VIM and VisualStudio. I use VIM because VisualStudio is slow. I have had a few Python programs where it is significantly faster to use VIM and python3 from the command line than to do the same in VisualStudio. I don't understand how Microsoft managed get python to run so slow!

QuickC is faster than VisualStudio too.

The context-sensitive help on some of those older compilers was better than what we have today.

QuickBasic in particular had a wonderful feature where pressing F1 would pull up the relevant help information. Then Microsoft went through a long period in which pressing F1 was likely to pull up help for a different platform (MFC, Windows, other languages) than what was needed. The C compilers were particularly bad, as MFC function calls had identical names to the underlying Win16 Windows calls. This carried over to OS/2, Win32, and .NET. It seemed that any help request involved wading through a billion irrelevant variations on the same call. The OpenFile, OpenFileA and OpenFileW thing didn't help either.

The incredible thing was that the ancient help facilities were off-line. They worked great without access to the Internet. This can be a big deal in a secure development environment, or in a factory, or in a remote location, or whenever developing without Internet access.

I have a distant relative with diabetes. One of the early indications was him getting disciplined for going to the bathroom to often. Not exactly sure this software will help ...

What happens when someone puts AI on the pitcher to track his mouth motion to detect when he is saying fastball vs curveball, etc?