Haha fair enough. If you do know someone at Instagram, please just tell them to search for me in their support inbox. I've included identity verification in my email. They could also just reach out to the email I had in my account before the hack occurred.
Great question! My phone number isn't particularly private, but I'm not sure how exactly they found it. I have a guess, but it's pretty elaborate, so they may have found a simpler route.
I agree with this, but feel like there might be issues I'm not thinking of. Can anyone shed some light on why browsers are so tolerant today and why that might be a good thing?
The Robustness Principle states "Be conservative in what you do, be liberal in what you accept from others."
Following that maxim, browser developers assumed that, even if HTML wasn't inherently correct, if they could figure out what the user logically meant then assuming that was better than not working at all.
In short, people wrote garbage HTML and it proved easier to fix browsers than people. At first, it wasn't too problematic, but as HTML got more complex more problems surfaced and now everything is a mess.
This was the goal of XHTML: HTML that was required to validate as XML or it wouldn't work at all, and some browsers were, indeed, strict at this. The idea was that you'd only use XHTML if you were generating it with an XML parser or some other template generator that could produce valid code. In reality, that just meant that browsers that didn't understand XHTML treated it like HTML and worked, and browsers that did understand XHTML and validated it would show errors. Thus, users saw that browser X (doing the right thing) couldn't display a site, but browser Y (doing the wrong thing) could.
Probably legacy reasons and the type of errors you can get.
Since JS used to be "sugar on top", it wouldn't make sense to completely eliminate the page when that piece of code which makes a logo flash doesn't work right.
Also, you can have JS errors coming from loads of places. What if an extension you use has a bug in it that triggers only on certain sites because of some stupid unicode issue? What if some ad has an issue like that?
And basically, it really boils down to: we all ship buggy fucking software. Everything has some kind of a threshold for errors (or errors that blow up only under certain conditions). It's good to have some built-in fault-tolerance that prevents an all-out disaster.
A company I worked for (property appraisal company) got acquired by another company. They forced one of the higher up developers out. 6 months later that developer wrote an app dealing with lines at Disney and was making money on ad revenue. The company came after him claiming they owned it because he used skills and knowledge he gained while employed to write that app. We had a "everything you do at any time belongs to us" contract.
He was at the original company for 15 years at the time and was the 2nd developer at the company. So he had a lot of friends at the company and he told his friends at the company that he made this cool app and word got around because everyone thought it was harmless to repeat.
They got acquired by another company and then gutted, most of the management that did that got thrown out after anyway so I don't know how much it matters.
That being said I always read my employment contracts now.
Google has an application process where, before you've written any code, you can ask for copywrite release. However, if you don't do that, they can and WILL claim ownership over your personal projects. It's happened to me.
Google is actually pretty good about this. They clarify some areas that are no-no, and there is a simple process for asking if a side project can be released as open source.
I bought display ads in Nature (the journal) a while back. There's no way to purchase without talking to a salesperson. However, I expect they want to keep it this way.
I sell advertising on my site without any automated ad network and it is fantastic.
Automating the process might be easier in some respects but I am very protective of of my readers and only want ads which are both high quality and relevant. Having to talk to me is a good filter to ensure that happens.
Why not? I’d agree that current models can’t, but what’s the fundamental shortcoming of deep learning?