Show newer
John BS boosted

Legendary book. For all young folk out there: start here!

RT @FrnkNlsn@twitter.com

A very unique textbook "Information theory, inference and learning algorithms" by Sir MacKay combining Information Theory with Machine Learning. Nicely written!
Book PDF freely available at: inference.org.uk/itila/book.ht

🐦🔗: twitter.com/FrnkNlsn/status/15

@alper Why not guix? It's 100% libre (with options for less-libre things like firefox if you want) it's build from source (unless you want to use substitute mirrors) and has a lot of nice features! The only downside is it's kinda tough to get answers when you get stuck on something, as there's not always an easy 1-1 correspondence between something like the arch-wiki and Guix's configuration method.

Really diving into common lisp to write code for my startup. It's pretty sick.

John BS boosted

With the recent announcement that #Stadia is shuttering, this comic is once again very relevant:
xkcd.com/488/

John BS boosted

All the bike lanes in the Netherlands
i'm so fucking jealous this is litterally the entire country

@trinsec I use fresh garlic too, but for certain things like pomme puree ("mashed taters" where I'm from) I prefer the granulated version because it's not as astringent, but preserves most of its savory notes.

There's also black garlic and roasted garlic, both excellent additions to make a compound butter with other herbs and such :blob3c:

@trinsec Have you tried granulated garlic (different from garlic powder)? It's excellent :D

@tiago

Thank you for your response!

I appreciate the clarity on the funding issue; however, my point was that despite not funding research, many private and public institutions *could* do so. In fact, Stanford already does this with a large portion of their endowment interest. Of course, they have $28.9B in endowment, but if we start trimming the fat on administration (or football coaches...), for example, that frees up more funds that *could* be allocated to better compensating researchers. [Example:](theatlantic.com/education/arch)

On to the societal service perspective; I'd like to understand how you would define the term. I find the EU examples you provided more agreeable/obvious (e.g. tuition free colleges), so it may be my experience with the US public/private sector jiggery-pokery that is clouding my evaluation here.

Finally, the real question I have is this: academics are well aware of the grueling hours, the comparatively low pay, and the almost monastic devotion required to achieve tenure. So what's the incentive/motivation, and why select academia over the alternative, if not for status? If the answer is truly a well-intentioned "I'm just passionate about it", I think the passion tax analogy applies well here, and it should be addressed.

For me, I absolutely love doing research, but a well-paid, 9-5, more comfortable, private research position at Google's DeepMind seems more palatable than the academic alternative, despite my love for the academy and the desire for freedom I know comes with it.

If you don't mind me asking, what was your motivation to pick one over the other? And if you could propose a solution, to the original post-doc (et-al) compensation conundrum what would it be?

Thanks again for the discussion, I'm looking forward to hearing to your perspective! 😄

@trinsec The "bigger on the inside" part or the "4-dimensional time-linked entity that allows all the snot from all points of the past and future to be draining out of your nose at the present" part? 😂

@tiago At the risk of starting a flame war (not my intention), I'm going to strongly disagree with the "misplaced comparison" notion here. Keep in mind, I'm mostly familiar with US based institutions, so please take this with a grain of salt. (Also, the original premise is a false dichotomy: "you can work at a company any not benefit society OR work in an academic position and do a public service to society". Most societal wealth generation is done by corporations, not to mention most QOL improvements over the last ~200-500 years were from primarily private enterprise, but that's a whole different rabbit hole...)

Being an academic and performing (publicly or privately funded) research while teaching, etc, at a university is not a service to society in any special sense. It is, frankly a job. A job paid by a (likely for-profit) university, typically with massive endowments (and huge income from undergraduate attendees). Lately, these universities have been finding increasingly asanine reasons to balloon their administration staff to justify skyrocketing tuition costs, despite no appreciable performance improvement to the student's education, nor an improvement in research/graduate student funding by the university. (I'm sure the parallels between big tech and universities are obvious at this point, so I won't belabor the point.)

For example, my university has an endowment of 3.8 Billion dollars. With a B, and that's a mid-sized public institution. The idea that there's "not enough money" to pay professors/research staff/postdocs/graduate students a decent wage for their highly trained skills and expertise is laughable at best, and a baldfaced lie at worst, especially when you consider their actual diversified income streams as well.

Of course, those on the tenure track can cope with their abysmal compensation to talent ratio by claiming they're "performing a public service" (and departments force their graduate students to cope since they have no real negotiating power). In reality, they just care more about status than money, so they either portray the "selfless martyr for a greater cause" trope to accumulating social capital, or they maximize their status-based payoff function in other ways, typically by becoming experts in their fields - even if that field has 0 actual utility (some pure mathematicians have even made this a point of pride). Or both, both is good.

If your goal is status over wealth, that's fine. But treating a class of jobs as worthy of less compensation because you view it as a "public duty/service" isn't rational, it's a passion tax that we have known exists for decades in fields like nursing, geriatric care, and teaching. Basically, passionate people are shafted because they'll deal with it due to their passion to work in the field, and the intelligent ones will find some way to justify it to themselves and others (as exemplified in the original tweet).

Instead of coercing passionate, talented, motivated, highly specialized people to accept subpar working conditions/compensation because they "just care so much", how about we actually pay people based on the amount of value they generate for the institution, rather than exploiting them for cheap labor? If they wise up (as many postdoc candidates have) then we will have a glut of PhDs and a dearth of filled Postdoc positions. Who knows how that'll pan out...

(I'll add sources for my claims in the morning, but I'm on mobile and it's 3am).

If you have time, please let me know what you think; I'd love to hear your perspective 😄 And sorry for the wall of text 😅

@jmw150 @lucifargundam Thank you for coming in clutch on this one 😭 I was struggling to find it elsewhere xD

Does anybody have experience with Disroot who's willing to provide their opinion?

Pricing seems very reasonable but I'm somewhat dubious of privacy claims, considering many "private" utilities have inevitably been compromised by government orders or bad behavior (e.g protonmail, tutanota, lavabit, etc.)

@jmw150 I may actually use this for my thesis if I can grok it, thank you so much for posting it!! 😄

@jmw150 How similar is this to a genetic programming approach? Are they leveraging neural networks? If so, how transparent are the resultant programs to analysis?

@trinsec Oh! I do woodworking regularly, I can chime in here for definition clarity:

Staining works by applying a dye to the inner fibers of the wood; it leaves the grain of the wood visible (and often accents it), and penetrates ~1-2mm into the wood itself (depending on grain porosity).

Paint is almost always applied after a primer coat (to prevent the wood from absorbing it and causing some areas to look less painted than others), and contains pigments, rather than dyes, and sits on top of the primer layer.

Both stains and paints require a sealant on top of them (or built into the product itself) that is typically a polymer such as poly-urethane, an epoxy resin, or some sort of "hardening" (read: polymerizing) oil such as tung oil, or urushiol lacquer (super high-end Japanese woodworking finish). Multiple finish coats are usually required, and sanding between coats (unless you're using a self-wetting lacquer) is almost always the best policy to prevent the coats from building up unevenly.

The final coat is what provides the desired amount of sheen to the surface, along with the finish-coat composition, and method of application. For example, shellac can achieve a very high gloss with fairly little effort via the "french-polish" technique, whereas an epoxy finish is much harder and will require a floor buffing pad and a series of polishing compounds to achieve the same gloss.

Also, in your particular case, the (at least American English) term would likely be staining, even if it's a commercial, 2-in-1 product that contains both a dye/stain and a finish compound for protection. Another good way to tell the difference is paint is almost always opaque in the can, whereas stain is translucent or looks kinda like a decarbonated soda.

Very interesting takes here! Someone tell me why he's wrong, or why we aren't doing better if he's right. Cleaner energy is always a good thing, right?

youtube.com/watch?v=4aUODXeAM-

@barefootstache I'll argue no: imagine the prompt is a function that accepts three boolean inputs which are evaluated only at runtime. After submitting your answers, the prompt would evaluate them, and at least one must be a lie. If you respond with a recursive function call to the prompt, you would also need to supply 2 other truths and a lie for the function call to evaluate to false, and then the original call to evaluate as false. However, you have no way of inputting these other three conditionals. If you did, you could flatten the game to 5 truths and a lie, since boolean expressions of this form would have to evaluate such that exactly 1 is false. So...really, you just have n truths and a lie if you do it this way, which is only valid for the original problem definition if n==2.

Thoughts?

Yeah, CommonLisp and Julia (which is Lisp based, funnily enough) are solidly my top languages of all time, and they should be yours too. Fight me 😂

I legitimately don't understand why people didn't just stick with it after it was developed in the 60s. It has so many excellent features, the syntax keeps getting cleaner due to macros, it's really fast, and the major complaint is just the parentheses? I'd trade the parens for interactive, stack-preserving debugging any day.

I'm hoping we can have a bit of a CommonLisp "revival" soon, as it's incredibly stable and just gets so many things *right*. The only downside is a lack of community/"political will" so-to-speak, to get people to get through the prickly intro and to be able to really see how excellent it is.

Show older
CleverLibre Social

CleverLibre Social is an inclusive social instance for open discussion, learning, and community.
All cultures welcome.
Hate speech and harassment strictly forbidden.