You are missing the entire point of the oral defense while almost stating it. The existence of the oral defense is what extrinsically motivates the supervisor to do their job well. If the student fails, the supervisor loses faces in front of their peers, which obviously they try to avoid. Removing the oral defense would massively reduce the quality of PhDs because supervisors will not do their job.
Also in my observational experience in physics, a fair number of students (probably like 10%) get some sort of major corrections to their thesis. After writing this, I did a search, and in the UK, the number across all fields is 16% [1].
The system is highly non-linear and has a lot of flippable components. For example, average temperature increase will cause the AMOC to collapse which will make parts of Europe cooler (while also increasing extreme weather events).
Similar things could happen with whichever glacier you are looking at.
The underlying problem is that the whole concept of cards is insane. Basically, you go around telling every shopkeeper the code for your safe, and ask them to take however much cash you owe them out of the safe.
Sane ways to organize payments:
- Merchant gives you a bill-id. You input it into your bank website - where you see the bill amount being charged. You accept, and bank pays merchant.
- You give merchant your card number (that's the only information - no expiry, no ccv, no name). A notification pops up on your bank website asking if you want to pay what the merchant is requesting. You accept.
- You go to your bank website and obtain a random number, either allowing a single transaction or a recurring transaction. You give the merchant the number. After merchant charges it, no other merchant can charge the same number.
This would also solve a major annoyance I have with card payments, which is no way to link a particular purchase back to a transaction.
With checks, you write a check and can write down the check number with a note about what the payment was for. When the payment posts, the check number is part of the transaction. With a card payment, they charge your card and, sometimes days later, there's a pre-authorization with some obscure transaction description. So many scary transaction descriptions that make me think "wait is this fraud".
With this method, the approval flow would also allow people to add a blurb for what the transaction is for.
> You go to your bank website and obtain a random number, either allowing a single transaction or a recurring transaction. You give the merchant the number. After merchant charges it, no other merchant can charge the same number.
That's basically how Blik works in Poland. With the exception being that the number is random 6 digits randomly generated when you open the app, that is active for ~2 minutes. So you don't deal with the issue of very long and error prone numbers to copy.
Cards have supported strong, positive cardholder authentication at the POS (chip and PIN) and online (3DS) since the 90s.
It’s entirely the US credit card industry and its regulating bodies’ fault that it has made neither mandatory in the way that e.g. the EU did, and is in fact fighting any attempt to do so tooth and nail (please think of the conversion rate!!)
Can you guys still write a check at the supermarket checkout? No pin credit cards are small potatoes next to personal checks in terms of weirdness. But I wish we had that trust culture everywhere.
He is explaining that if you design an app that displays as desired on the outer display (in portrait mode). Then if you display that app to the inner display (also in portrait mode), it will show up exactly the same, just with everything twice as large.
This obviously makes life easier for both users and developers.
There are two stages to learning research level math. First you learn to reprove other people work. This is relatively easy because the language, notion and prior explanations have already been optimized to help prove the next thing, and you know a solution exists. Then in your PhD you try to solve problems you don't really know a solution exists, and if it does, how long and complicated it is, what techniques will be used, etc.
You need to solve the second to get a PhD. For good reason. The second is way harder than the first. And now AI is making second way obsolete. It's not the end of the world, but it is the end of how things have been done for centuries.
Shops and factories can be temporarily shut down if they don't comply with regulations. Businesses that require licenses like law practices may similarly have to stop operations. That is half of going to jail.
Unfortunately, the growth potential that software industry generally provides has generally prevented governments for creating and enforcing similar laws for software products and such.
In the current system, political parties who do things that stop the economy growing are usually voted out, more than political parties that refuse to take action against nebulous criminal activity. That's why, for example, my city refuses to do anything to stop the factories in the city from dumping waste into the river.
Do we know if Google is operating on the frontier of the trade-off curve? What do we lose from Pixel Android when GrapheneOS security changes are made?
Very little if anything at all. Depends a bit how you draw the line between an app and the base OS.
Contactless payments dont work with Google Pay - but that is due to Google using attestatiom to check if the OS is GMS lincesed (GrapheneOS is not).
There are alternative contactless payment options that work fine.
There's quite a list you can get even from GrapheneOS's posts - among other things, there's a rather large set of apps that will outright crash when MTE is enabled.
If you are building your own harness that does correct citations, is the correct thing to give AI access to some deterministic tool that allows it to actually copy paste parts of documents its reading (with links), rather than stochastic reproduction that they do by default?
I did something similar for structured text extraction. I added markers throughout each source document and then, for each piece of info I wanted, I asked the LLM to provide two separate fields:
xyz
xyz_citation
The latter was just the node number. So then my code could extract the exact snippet, instead of trusting the LLM to quote something verbatim.
That’s what I did when I built my stuff. I have deterministic content with AI commentary, where it seems most people are doing this crazy thing of sending data through the model. I can’t understand it.
Any given single agent is not long-lived due to limited context. What impact does it have on the status signals the agent develop compared to the signals humans (who usually have much longer context) have developed?
Yes, they do store the value. There was a blogpost posted to HN last year that discusses the cryptographic features of these cards [1,2]. It's amazing that these cards are almost solving the problem of secure digital money.
Also in my observational experience in physics, a fair number of students (probably like 10%) get some sort of major corrections to their thesis. After writing this, I did a search, and in the UK, the number across all fields is 16% [1].
[1] https://www.lexacademic.com/blog/how-common-is-passing-with-...
reply