I think it will be challenging. The factors that lead to the current culture is so ingrained in the researchers. We are all kind of nitpicky and generally looking for reasons to reject papers.
Wonder it if would be helpful to put a page limit on the evaluation section?
Definitely not true in every field. But in PL, it is so rare to get scooped. I’m fine even having code on GitHub before submission.
Pretty much. Basically hard to hide all the overheads. And you can get really pathological situations with aborts. The last one means that probably locks are just a better option.
For non functional languages, I think people never got the overheads low enough across all workloads. And it had lots of pathological performance cases.
For HW, every Intel implementation had bugs such that they disabled HTM with a microcode update.
My reading is this cancels essentially all non-AI, non-quantum CS research. nsf-gov-resources.nsf.gov/files/00-NSF...
This is absolutely insane and lacks imagination of what AI systems could do. What if someone builds automated decision systems that can apply lethal force?
I’d apply broadly. Maybe someone has some funds for the short term and can’t make a long term commitment to a grad student. Maybe also consider industry as I expect faculty hiring will be challenging for some time.
We have tuition as a direct charge for graduate student researchers. It is fictional anyways. So it is effectively an overhead, but one that doesn't require negotiating with the government.
Wonder if we will see large tuition increases for grad students?
Mine hasn’t mentioned it yet. Submitting proposals next week seems problematic.
Yikes. Hope I can graduate my students.
I think today's recovery has to do with lots of puts needing to be exercised and shorts needing to be covered. It seems to have gone away. But this is all very confusing figuring out how to react optimally.
When do you think we will hit rock bottom? I pulled largely out of the market, so I have to figure out when to return. I unfortunately did sell some put options yesterday that I had bought on SPY as insurance before I decided to just sell out.
Yes, but is the market for that very large?
I guess I should add that I’m aware of the square root speed up for Grover’s algorithm, but I suspect constant factors will dominate that speed up for a long time.
I don’t get the industry interest in quantum computing. What useful thing could I do if I had one? Well, assuming I’m not interested in breaking crypto…
Came really close to getting a hybrid, but realized that a hybrid didn’t get us much gain for highway miles. The RAV4 prime didn’t fit us…
Unfortunately no. Wanted a vehicle for road trips to see more of California. Chargers seemed a bit of a problem still. Went with another Subaru Forester.
I would have bought a Tesla this month if Elon was the typical CEO.
It gets better.
If you have a good strategy, let the rest of us know.
I’m not sure what the right thing to do is. Cash has the risk that inflationary policies could eat it up, whereas maybe we would get some growth in stocks due to inflation?
I think we may be unlikely to see indirect cost changes to NSF. The reason being is that NSF grants have a total budget limit and not a direct cost limit. So if you drop the NSF IDC percentage, the PI just gets more money for direct costs.
How are you figuring out what activities are still safe? Is encouraging undergraduate research still okay?
If our goal is AGI in a form similar to people, it seems to me that queries I make and their responses should change the weights.
You should ask the SIMPLEST strategic questions that they will answer incorrectly and that they need to know to succeed at the job.
Would the result change if Canada got to divide itself into states such that the average area of a new Canadian state matched the average area of a current US state?
One other random thought. I guess it is possible for AI to get to human levels of ability, but at a higher cost. Like our generation’s Concorde.
