PhD candidates in the US usually get somewhere between $25K and $50K stipend, also some level of benefits (typically health care). Sometimes there is a tuition waiver (student does not need to pay grad program tuition).
In my case I was making $32K/year with a tuition waiver and health benefits around ~2000, in SF, which was barely enough to rent a shared apartment and eat food. The only way I could rationalize it is that I was maximizing my future freedom (job choice).
Yes, I suppose you could try to justify it as "this is the price you pay for having the freedom to build your own research plan in the future" ("maximizing your future freedom") but in reality, this just sets you up for more of the same- getting a faculty position (pre-tenure) with a low salary, and immense pressure to bring in funds via research grants/publish papers.
I eventually tired of the process and moved to industry because the struggle wasn't worth it.
No it's typical. It's just that your stipend is usually just x amount of Dollars + whatever tuition is so you never have to care what it is and you don't pay it directly per se, it's just included in your stipend. Someone pays for it at the end of the day though.
Yeah, I distinctionly remember a postdoc I knew who was irrationally excited to move to a role where they were going to get paid $35k, in 2010s money, and they were damn excited about it. And they were moving to a high cost of living area (from a high cost of living area). I was utterly flabbergasted because they were very smart, very technical and should have been earning 5-10x that. I feel like they didn't know what they were worth and academia had utterly failed to teach them that.
I don't know how they paid any of their accumulated (I assume) student debt, let alone had an even decent standard of living.
Those are typically skills a starting scientist needs to learn. At the same time, sometimes it does feel abusive especially if the student doesn't get some sort of credit for doing the peer review and talk prep.
In my program the main reasons people took a long time to graduate was: by year 6 you are usually very well-trained and highly productive (making you very useful to your advisor), and advisors often require you to publish an important paper in a major journal (Science, Nature, Cell) before they sign your dissertation.
Scattering has long been key to making realistic looking rendered images. One of my favorite papers: http://www.graphics.stanford.edu/papers/bssrdf/bssrdf.pdf
which I think is the first time I learned that rendering milk is a tricky problem.
Several tips helped me move from "painting with solder" to "hmm, that's acceptable": "heat the component, not the solder", "taping things to the table saves a hand", "use an analog, not digital, soldering iron", "clean your tip clean". Those, combined with practice, mean that I can do basic electronics work. I still accidentally melt insulation, and damage things from time to time.
Switching from a Weller to a Pinecil was also pretty nice although I'm sure everything I do, I could do with my analog weller.
AI models moved beyond next word predictors recently. Considering them to just be partially decent auto complete is completely missing many recent innovations.
I don't have a Bambu; previously, I had Prusa printers from the MK3 generation and I struggled to get good prints (poor bed adhesion and the extruder breaking frequently, requiring very intensive repairs); since not having a working printer slows my hobbies down, I ended up with two. Both broke down and I got tired of fixing them, but when I looked at the prices of new Prusa, they were high enough to make me pause.
Instead of a Bambu, I got a Flashforge Adventurer 5M. It is incredibly cheap (cheap enough that I am more than happy to replace it after two years if it stops working), and is pretty reliable (compared to the Prusa MK3 and MK3S I had), and most of all, the self-calibration works well enough that I don't spend any time debugging prints that fail at the first layer anymore; I just re-run calibration and it's fine, and if it's not fine, I clean the plate and it works.
It also comes with a terrible slicer (dervied from Slic3r I believe) with annoying "log into the cloud every time you start the app", but I moved to OrcaSlicer. I had to give up a few nice features but it hasn't truly impacted my workflow. And it does receive firmware updates (it's connected via wifi to my home network). My hope- just a hope- is that they don't do anything truly stupid with future firmware updates or end up getting in a hissy fit with prominent youtubers.
I agree. I would almost say the best argument to make is "liberal humanism is compatible with machine intelligence and regulated capitalism, but we must still remain wary that authoritarians will abuse the system".
I do see prices when I look at the instance selection box. "On-demand linux base pricing: 0.0104 USD per hour" for t3.micro (matching the published pricing). it does not show the full price (based on any additional volumes or other configuration details).
It gets far more complicated when you have reserved instances, and combine reserved instances with RAM sharing when working in a larger org.
Modern US coke doesn't taste much like the coke I drank growing up (late 70s, early 80s, before they switched over). I remember drinking "a perfect coke" on a hot day, it tasted almost "botanical". These days, the closest thing I can find is Mexican Coke (which they sell at Costco), it's a lot dryer (less sweet) tasting to me than US coke.
In my case I was making $32K/year with a tuition waiver and health benefits around ~2000, in SF, which was barely enough to rent a shared apartment and eat food. The only way I could rationalize it is that I was maximizing my future freedom (job choice).
reply