I am really glad arXiv is getting more funding. It is an essential resource.
For me personally, it’ll be really interesting to see if the frontend changes. No doubt there are some important improvements that can be made (a website can always be made more accessible, moderation tools, support for name changes as mentioned, etc.) However, to first approximation, arXiV’s website already seems almost like a platonic ideal. It reminds me of Craigslist. Simple HTML, loads fast, has the information and features you need but otherwise gets out of your way. I love it.
The arXiv team deserves a lot of credit for what they’ve done to get it this far. It’s difficult to overstate how useful and transformative preprint servers have been to science.
I had this same thought but was reluctatnt to state it as it feels like unnecessary pessimism. But honestly, this website works flawlessly. The last thing it needs is software developers trying to keep themselves entertained or impress people. If it ain't broke...
Cool idea! Is there a way to go to the arxiv page (https://arxiv.org/abs/...) of a paper instead of only going to the PDF (https://arxiv.org/pdf/...) without manually manipulating the URL?
Awesome, hadn’t seen that. I’m sure there will be a zillion edge cases where it won’t work properly for specific documents because of odd latex quirks.
If the pdfs were served by some of the megacaps who posts tons of papers (e.g. Google or Facebook) then it would be an order of magnitude faster. And said megacaps would end up spending peanuts relative to the value they get from arXiv.
An order of magnitude faster sounds unimportant. Is all the time spent by researchers sitting around waiting for 0.4s for a paper to download really an issue?
They probably could make it faster by rendering server-side a preview of sorts. I don't think that most papers are large enough for this to have a major impact unless you have a very slow connection.
SumatraPDF on Windows is lightning fast with the kind of PDFs you get from the arXiv (true PDFs without heavy graphics). I don't know what is fast on Linux, but probably some MuPDF-based viewers are (the MuPDF PoC viewer is fast even on WSL).
Some people are angry that arXiv which is probably the most important repository of science are getting public funding because big tech companies are using it to publish their work, and they should fund that from their deep buckets. Are you aware of the consequences of what you are suggesting?
If you are funding something, you have some authority over what you are funding. Directly or even indirectly (If you don't follow our suggestions, we might revisit our financial contribution). So you want to put the place that should be open and independent under the thump of corporations that does not always have the same goals as the benefit of the science and public interest?
There is nothing wrong with chucking money at Archive to help with costs, provided that money does not come with menaces. The basic idea is that if you profit in some way from an open repository of ... something then why not contribute to it? If enough people/orgs contribute then we are all good. Otherwise, if the org behind the project is an individual doing it as a hobby and they can't afford to continue then they might need support.
In the end Archive and co have costs attached. You don't have to contribute but it might be nice if you did.
Public money indirectly comes from the most profitable companies, so it's not as bad as you suggest. I'd also say there are worse freeriders of academic research in the economy than big tech.
There's sales tax for customers, income (and payroll) tax for employees, and then taxes on dividends, and then capital gains on stocks. Low-margin SMEs contribute proportionately very little tax compared to Google when we account for tax contribution this way.
I guess you're referring to my specification of low-margin. As I also specified SME, what that means is both low-margin and low-revenue. And the true intented meaning is that all cash flows one might want to tax (mostly with progressive rates) are small.
> And the true intented meaning is that all cash flows one might want to tax (mostly with progressive rates) are small.
rates become high fast enough for employees of sme to start paying substantial taxes, and payroll taxes don't have progressive rates, and ss is actually regressive tax.
Like I said, it's very rough since the transition happened over several year. I dunno, you can eyeball this plot; if anything I'd say it didn't become default in CS until after 2010:
Prediction: Arxiv has no need of this money, as it has no need of upgrades. This money will exclusively attract salesman, all of whom will offer to make arxiv worse, for money. One of them might succeed
I concur. Seriously it's job is to act as a repository of PDF files. It does that it does that well. It doesn't not need a fancy new frontend that changes again in 18 months because "circles are in" it doesn't need to move to a "distributed load balanced fault tolerant web scale platform solution" (that costs about 2x in maintenance and engineering time as it currently does).
It doesn't need to do anything of these things. It needs to serve documents and thats it. If there are problems with load they can always throw it behind a CDN since all of their contentt is largely static it fits nicely into the use case CDNs were developed to solve.
Oh great, looking forward to a captcha every time I download a paper.
(FWIW, arXiv is one place I trust not to make such stupid decisions, but migrating to a cloud often means the decisions are no longer made in-house...)
It's possible that they jump straight to the hybrid solution that many companies now make. Like migrating to the cloud could be just to have backups of papers in s3 with failures to download them initially then pass through to the s3 location seamlessly. It could also be trying to rewrite everything to run on lambda.
My point being that migrating to the cloud means different things to different companies/groups.
I suspect this cloud move is more for security -- to keep others out -- than to improve the product. As you mention, many decisions will no longer belong to Cornell. This is indicative of larger geo-political trends.
Popping arXiv would be quite an impressive watering hole attack. You wouldn't need to do anything to the site itself, their PDFs are downloaded by a huge number of bleeding edge researchers and technical journalists.
Arxiv is a blessing, I would not be where I am without Arxiv and all the folks openly sharing their research.
I don’t like that deepmind gates their research behind nature and other publications. I acknowledge it puts them in the cool kids club but it feels unauthentic. Putting research out in the open moves humanity forward.
Arxiv can say it made a little dent in the universe and I’m glad it’s getting the funding it needs to be sustainable.
I just hope they don’t burn it out trying to scale needlessly.
Oh that’s great to hear! I really like arXiv and am glad it’s getting more funding! I hope this will enable arXiv to fix some of the long-time issues it has, including the way identity and attribution are handled.
For example: the current system assumes people’s names do not change [1], which in particular negatively affects trans authors. This seems like it could be fixed if they had the bandwidth to, and I hope they will.
There is no good way to deal with name changes. You can change them in the metadata, but the old ones will still be around in the paper itself. You can let authors change them in the paper, but the old ones will still be around in others' bibliographies. You can't do much about the bibliographies, since (1) you'd need an automated way to find them all and disambiguate them (have fun with Chinese names) and (2) a lot of authors will reject the idea that someone else changes their papers.
For disambiguation there's ORCID, and for the citation there is nowadays the DOI to identify the paper. So long as you can update whatever the DOI points out with a note it seems like this is handled?
OK, so I typically use DOIs in my bibliographies, although I seem to be in the minority for now. What next? You have to find the exact name of the author that changed and edit it. Let's say you do so (not a trivial undertaking since references do not come in a standardized format even if the DOIs do; by the way, arXiv doesn't understand bibtex). Now my paper says "Merton and Fox observed in [5] that...", and you know that Fox (one of the authors of the [5]) became Wolf. Will you change my text to "Merton and Wolf observed in [5] that..."? How can you tell that this particular "Fox" is a surname rather than the animal? What about acknowledgments, which come with no DOI or ORCID?
You make a good point that one should use the newish \orcidlink facility whenever one first bothers to type out the name of an author.
It doesn't seem quite as critical when the name is only part of the citation, so long as the ORCiD is in author list at the other side of the DOI.
But I think you misunderstand me regarding who should do the work: the original author is in charge of ensuring their ORCiD is included in the author list and that its entry is kept updated.
If you are just citing a work, just give the doi and whatever authorlist you found at the time.
Then I'd say that if you want to refer to a specific author of a work, you need to lookup their ORCiD and use whatever they specify as their preferred name. Also include the ORCiD itself so that the next researcher to come along knows where to find the up to date name.
This is more trouble than it's worth for the mathematicians of today. What exactly is gained from these ORCIDs? A DOI auto-generates a clickable link, allows for automatic reference tracking, and (hardly an intended feature but perhaps the best one) helps readers find a copy in places like sci-hub and LibGen (on sci-hub, DOI is the only lookup option that currently works). With an ORCID, I could make it double-clear that the D. E. Knuth I'm citing is the real Donald Ervin Knuth. Nice, but I'm supposed to learn a new command and spend 10 minutes researching authors per paper for that?
The standard convention in citation is to follow the format:
FI, MI, Last. For instance, if an individual changes their last name from Davidson to Mitchell, the change would manifest as follows:
Suppose an author named Susan Thomas Davidson (S.T. Davidson) decides to change her last name, becoming Susan Thomas Mitchell. The citation would then be updated to (S.T. Mitchell).
This alteration retains the consistent impact as the last name, serving as the clear identifier, undergoes the change."
It seems arXiv does support name changes, just not in citations and references[1]
This seems… fine? To modify citations would be a strange alteration of history and an intrusion on authorship.
If someone quotes this comment and cites it as “bigyikes” and later Dang changes my name, should he also edit the comments that quote me? I don’t think so, but maybe I’m in a minority here.
The link you shared seems great, thank you! I was talking to a friend just yesterday who regularly publishes to arXiv, and my understanding was that changing author names on papers where they weren’t the primary author wasn’t possible. But from the link you’ve shared it seems they’ve made the steps needed to address that. I think this means I might be able to deliver some good news!
arXiv is a great resource. At least in my field (Programming Languages, but I imagine it applies to most if not all of Computer Science) it seems like most papers are on arXiv.
I see no reason to not make papers free. Most things cost money because either they're material themselves, or cost money/material to produce which needs to be somehow recouped. But papers cost almost $0 to distribute and are created with grants which get funded by the results and discoveries themselves.
My philosophy is that if people want to spend (waste) time on closed journals more power to them.
Potentially when fields were small and niche this aggregation of topics might be useful. Today in modern search and computerized tools, the question is: why do we need them?
There's nothing in academia that I feel is more of a racket than this scheme.
I feel that _everything_ should be published and that natural discussion of said topic will drive the importance.
Today, with modern search and computerized tools, one of the primary heuristics I use for determining which preprints to read is "do I know any of the authors". If I don't know you, your ideas are most likely not worth reading, even if they appear to be about a topic I'm interested in. There are so many papers published every year and so little time to read them properly, and the search tools that exist are not that good. If you don't have an established reputation, your best bet is submitting your work to a relevant journal or conference, where the peer reviewers will usually give you a fair chance.
Google Scholar is a good example of how bad the tools are. Its recommendations are apparently based on things like the papers you publish, the papers you cite, and the papers that cite your work, which sounds superficially reasonable. That worked well enough when I was doing my PhD within the boundaries of an established academic field. But when I started doing more interdisciplinary work, the recommendations quickly turned into garbage. A better recommendation system would understand that there is a field I work in, an upstream field I take ideas from, a downstream field I contribute to, and several fields further downstream that use the work I contributed to.
I feel this is already true of books, the question is will a publisher put it out... books cost money to produce and by comparision to hosting are massive investments.
Actually, yeah. I think it could make a difference. Usually there is some basic spell-check, grammar check stuff already going on, but the kinds of things that would be easily fixed/identified by LLMs could be:
- continuity checking - often moving sections of text around introduces temporal errors
- style cleanup
- quasi-duplicated text that came from multiple edits or start/stops being merged
- fixing vague and unclear passages
But probably what they won't do well in a near terms is provide more critical feedback, such as when to throw away large sections of book, or remove subplots, or where to dive deeper on other topics. I'm sure they can generate general suggestions for this kind of thing though.
> In addition, arXiv will provide substantially better access to the visually impaired by producing HTML as well as PDF versions of its content.
But it's a different question whether they will succeed. A compile-to-HTML pipeline needs to work with every single LaTeX package that people use, or at least work with most ones and fail gracefully for the rest. That was hard enough for compile-to-PDF, requiring the arXiv to keep every year's version of TeX Live permanently maintained.
I hope they just use the money for more bandwidth and maybe spend it on free api/download access so people can get big chunks of it easily and build all kinds of things (from alert on keyword to just stream of text to finetune models)
And I hope they don't touch the frontend at all :)
I wonder if they could use some of it to make an experimental post-publication review system. I'm not sure what the future of peer review is, but I doubt it would hurt for them to try.
I think it's antithetical to the philosophy of arxiv in the first place. It's a preprint service. Adding a review process would make it... a journal. Arxiv being impartial to the content it publishes is one of its' greatest strengths.
At a higher level of abstraction, arxiv is an attempt to fix problems with the review process. Attempting to fix more problems with the review process would.... Etc
Don't you find it just a bit mysterious how a site that already works perfectly well and does a technically pretty basic task can spend that kind of money?
* overhead - people cost a lot of money: this includes ongoing maintenance of the tech stack, including rewrites of obsolete parts, but also ops cost - preprinting can include more than one human touchpoint
* converting PDFs to HTML is an annoying problem
* searchability of the repo is likely an annoying problem
* any new features that stakeholders want added (commenting, annotations, etc)
* ongoing hosting / CDN cost
How much do each of those cost and how does it add up to 10M? If you get two overqualified people to work on it full time and pay them FAANG salaries it'll still be enough for decades. I can't imagine the hosting is expensive when 99% of the papers are a few megs at most.
I'm just a bit confused because this is a site that already works really well and isn't technically difficult.
Eh, inflation has happened, and taxes, medical bills, and real estate are insane these days. So much so that you can't live a "normal" life of the standards of 20 years ago with less than $500K+ in salary. (At normal "middle class" modest levels of life hat's $250K after taxes, $200K after medical expenses, $150K after rent, $100K after food, $50K after transportation, $25K after food ...)
Sounds outrageous but most people have been brainwashed into thinking living with roommates in a cramped flat full of mildew, giving up all dreams of owning a home, and driving an old, unsafe car is the new norm.
Given that, $10M isn't really that much, considering you need to hire software devs to do any of the work they propose, and those software devs are getting $500K+ offers everywhere else.
> At normal "middle class" modest levels of life hat's $250K after taxes, $200K after medical expenses, $150K after rent, $100K after food, $50K after transportation, $25K after food ...)
99% of the country doesn't pay $50K per year in rent. Nor $50K per year in food. Nor $50K per year in transportation. Nor an additional $25K per year in food.
For example they think they pay 2K/mo in rent but that easily blows up to 4K after hidden costs that aren't on the sticker price of the lease.
Utilities, utilities of others that landlord charges to you because the lease said they could (and so does every other lease in a 100 mile radius), maintainence and safety issues that landlord doesn't cover, moving costs averaged over time, extra months of rent you have to pay because of stupid 12 month lease systems, tax or whatever other stupid fees charged by the rent payment systems, pest control, mold control, cleaning costs, furniture costs, ... only after all that does the place become even livable for a middle class standard.
Yeah, if you're okay with some pests and bedbug-ridden furniture you can save $200/month. If you're okay with shitty internet access that you can't even have a quality video call, you can save $50/month. etc. If you breathe mold all day, you can save $50/month. If you move all your stuff with muscles and no moving companies, you can save $200/month. But then you're not having a middle class life.
Sorry buddy. I track my finances - since 2007. I rarely look at them, but I have all the data for me. And I'm not frugal.
Do you have yours?
> For example they think they pay 2K/mo in rent but that easily blows up to 4K after hidden costs that aren't on the sticker price of the lease.
This may drive you nuts, but in every place I've rented, I've never paid water, trash, sewage or parking. And I did not seek out these places. I know places do charge for these, but they are locale dependent. In a number of apartments I didn't pay for heat either.
Moving costs? Only paid them once. Otherwise just moved with friends, or had the company pay. OK, I did have to pay U-Haul. You got me there.
No hidden fees. Sorry.
As I said, there's a whole country you haven't seen.
My landlord charges me $60/month for water/sewer. I pay about $120 for electric and gas combined most months. This is in an area of the country where it gets above 100F in the summer and can easily dip below 15F in the winter.
> moving costs
Stop moving so much? I know very few middle class people that would pay a moving company. Renting a truck and loading it yourself, or using one of the several pod type services is the norm for most middle classers. I certainly wouldn't qualify this as rent. A moving company is a luxury purchase. Ideally if you have to move for work, your company would pay for this if needed.
> tax
Eh? There is no tax on rent. Your landlord doubtless has to pay property tax, but they would have to make that up via the rent.
> extra months of rent you have to pay because of stupid 12 month lease systems
Once again, stop moving so much. A lease is a protection against your landlord raising the rent on you. If you don't like it, then go month-to-month and pay more.
> pest control, mold control
Depends on the state, but landlords are generally legally responsible for mold. It sounds like you had a specific issue with mold, because you keep talking about mold problems. There is almost always an underlying cause of mold that should be dealt with directly.
> cleaning costs
I do not expect my landlord to clean my toilet. If you are hiring a maid, you are by definition not middle class anymore. Middle class people scrub their own toilets. No, really, they do. Well, some of them don't, but that's a different problem.
> pests and bedbug-ridden furniture
I'm not okay with pests, and my total furniture cost did not come close to doubling my rent. I also don't buy new furniture every year, and I'm okay with using a bookshelf that is 5 years old.
Most middle class people are not frequently purchasing new furniture, and if they are, they are going into debt really fast.
> mold
Again with the mold. We're not all breathing in mold, okay? Paying $50/month for "mold" makes no sense. How much bleach are you buying with that? It sounds like you have a water leak.
> shitty internet access
Sure, get good internet, but that's still not going to double your costs.
> no moving companies
Yeah, that is how middle class people move. We rent a truck or a pod thing and ask nicely for our family and friends to help load the coaches. If you're moving for work, the company gives you a signing bonus to pay for moving costs. Once again, stop moving so much.
Adding everything up in your post, I suspect you live in a very high cost of living area with a limited amount of old generally poor quality housing stock in a humid area (mold mold mold), probably the Bay Area. If you choose to live in the Bay, then your experience has very little correlation with what the rest of the country is like.
At any rate, if you earn $500k/year, then you are by definition not middle class. That's probably why most middle-class people don't live in the Bay.
> Again with the mold. We're not all breathing in mold, okay? Paying $50/month for "mold" makes no sense. How much bleach are you buying with that? It sounds like you have a water leak.
Why do you think I moved?
> We rent a truck or a pod thing and ask nicely for our family and friends to help load the coaches.
That works in the outback but where I live friends don't ask friends to move. Many of my friends have kids and are overworked, nobody has time.
> pod thing
My new property manager banned pods
> We rent a truck
Yep, I did this. Rented a U-Haul 3 times over 3 weekends. Wasn't cheap, AND tiring. I moved most of my own stuff, and it cost me close to $600 for everything, including the U-Hauls, the Uber rides to/from U-Haul, etc. And then another $400 for a couple guys to move a couple furniture items I couldn't move on my own.
You do realize that's $250K after taxes, here in California?
Which is $140K in year 2000 dollars.
Houses here cost $1M for a dump, or $2M for something decent. 20-30 years ago, being able to afford a home was a pretty standard thing for anyone in the middle class.
A 1-bedroom rental apartment costs $2500/month for something moldy and not wired to modern electrical code, or $3000/month for something clean, safe, and sanitary (you pick). Plus $500 for basic utilities (water+gas+electric+internet+phone).
No, seriously, it's insane that you're trying to defend this claim.
Median US household income after taxes in 2022 was $64,240. If you make *quadruple* median household income after taxes, you are not middle class.
Median post-tax household income was $34,123 in 2000, so $140k was quadruple that too.
Median pre-tax household income in San Francisco is $126k.
The standard guideline for an apartment being affordable is "30% of your pre-tax income or less"; a $3k apartment would be affordable by that metric for someone making $120k before taxes.
In San Francisco itself, single-family houses do start at around $1M, but it is not remotely true that you need to spend $2M for "something decent" - there are tons for like $1.2M, I'm looking at Zillow right now. You can get a two-bedroom condo in a pretty good neighborhood for like $750k.
House price affordability guidelines are "3-5x pre-tax annual income", so even in SF, the paradigmatic housing-crisis city, you could afford a $1.2M house on $250k or a $750k condo on $150k.
All of this is just setting aside the question of "if you can just afford to buy a house in one of the most expensive places in the country, does that make you middle class", for which the actual answer is no.
But by absolutely no sane metric is $500k/yr "squarely middle class".
That makes sense in a world where taxes are 15%. But in a world where taxes are close to 50%, you probably want to be thinking about 30% of POST-tax income, not pre-tax.
If you rented at 30% of pre-tax, take away 45% for taxes, you have only 25% left. There's no way you'll live on 25% AND have room for savings for purchasing housing.
> House price affordability guidelines are "3-5x pre-tax annual income"
Uh, no, I'd say 3-5X post-tax post-food post-rent post-everything-else. That would mean you could afford a house in 3-5 years.
Seriously, I live here and almost nobody with a total household income of less than 600-800K actually owns a home here.
Not sure I agree that that's middle class levels of expense. If you're living a "normal" life, and for the sake of the example you start with $500k per year, my estimates are $300k after taxes, $290k after medical/insurance, $240k after rent/mortgage, $220k after food, $200k after transportation. IMO $50k for rent/mortgage is the only vaguely plausible number you have.
- health insurance rarely pays for what they do, balance bill you several thousands of dollars every time you have a regular check up for an incurable health issue, and it often requires legal notices and lots of lost work hours on the phone with debt collectors, insurance, hospitals
- rent figures don't include my $200 electric bill, $200 water bill, $70 internet bill that comcast decided to charge me $170 last month out of nowhere, $50 trash bill, another $200 in maintainence just to keep the place livable because the landlord won't do it for you
- transportation isn't cheap. a rock hit the bottom of my car last month on the highway and it cost $700 to repair. tires wore out and that costs $1000. DMV charge me $600 a year just for the hell of it. insurance costs $200/month. other costs are upwards of $200/month just to use the damn thing. and then there's parking, energy costs, fastrak costs, ...
> health insurance rarely pays for what they do, balance bill you several thousands of dollars every time you have a regular check up for an incurable health issue, and it often requires legal notices and lots of lost work hours on the phone with debt collectors, insurance, hospitals
Despite what the news will tell you, most SW engineers don't deal with this. In over a decade of working, with lots of interactions with the medical system, not one of these has manifested for me nor for my friends.
> transportation isn't cheap. a rock hit the bottom of my car last month on the highway and it cost $700 to repair. tires wore out and that costs $1000. DMV charge me $600 a year just for the hell of it. insurance costs $200/month. other costs are upwards of $200/month just to use the damn thing. and then there's parking, energy costs, fastrak costs, ...
There's a whole country out there where you don't pay most of these fees. I pay a sixth of what you do to the DMV, and half what you pay for insurance (and I have more than the bare minimum). Repairs? Yeah, sometimes I have to pay a lot. But something on the order of $700 for a single repair is one of those "once every 3-4 years" things for me, for 2 cars.
> Despite what the news will tell you, most SW engineers don't deal with this. In over a decade of working, with lots of interactions with the medical system, not one of these has manifested for me nor for my friends.
I'm a software engineer, and I dealt with this multiple times a year until I joined a big corp. In all my years of startup life it was the norm to constantly battle the system and have several thousand dollar medical bills thrown at me all the time even though I had insurance. I don't actually talk about it a lot, because there's nobody to talk to who will actually fix the system, short of moving to another country. As I understand it, this is the state of "21st century life in the US".
At a big corp, things are different, insurances are shit scared of big corp HR ditching their policy so they pay for everything.
I mean this with the utmost sincerity: Get better insurance. Or work in a company with good contracts with insurance companies. Like probably any Fortune 500 company.
(Either that or understand your insurance better).
Back when I was a contractor I was on $850/month insurance and they paid for almost nothing. Deductible was $6K and even after that they paid for almost nothing.
Maybe $1600/month+ insurance would pay for something, but then it's questionable whether it's a good deal.
None of the marketplace plans covered the only couple of medical institutions in my area I trusted to give me a life-saving medical implant for a condition not well understood.
That sucks. I can believe if you have a tough medical condition then there's no limit to how expensive life can be :-(
And yes, marketplace insurance is poor. I get paid a fraction of $500K and the deductible for the whole family is under $4K. Max out of pocket is about $5K, but we never hit it. I don't pay any premiums. If I were single, it would be much cheaper - I mostly pay for the dependents. Big companies negotiate good deals, and they'll often have employees who will act as your advocate for weird billing problems. Insurance companies don't want to upset a big client.
Sad fact of life in the US that one of the reasons people cling to soul sucking jobs is this - even when they have enough money to retire (but still too young for Medicare).
Arxiv is run by Cornell, which is in Ithaca NY. Upstate NY a $500k a year salary would literally buy you a mansion. Ithaca right by the college is pretty expensive by upstate standards because of houses being converted to multi-family units to rent. But Ithaca for $500k is buying you a big house walking distance to downtown, if you go out of the city $500k is buying you a big house with a large amount of land.
But the government will steal $250K of it right off the bat. And then probably another $50K because at least in CA you get double-taxed both when earning and when spending.
10.25 percent (highest sales tax rate in CA - 1) of 250k is closer to 31.25k. State and fed taxes income taxes are around 45% but I'll grant you there's other hidden fees (social security and others)... luckily hidden fees are now illegal in CA.
You can easily get a mortgage for 3x income and 4x isn't that hard. So $500k can buy a $1.5m house with no problem. $500k income is easy to buy a house forth $500k cash after a couple years.
I'm appalled at the perspective of middle class for you. If $500K is barely middle class, how would you classify a grad student? I suppose you were once a grad student too.
I was. But (a) a lot of things are subsidized as a grad student (my dorm room rent was 30% of market price for the area), and (b) savings are nearly zero. It's close to a paycheck-to-paycheck life unless you have some external income (which I did, via Google Ads that I slapped on a 250K MAU/month website of mine). That external income paid for me to eat healthy, buy outdoor exercise equipment, have a couple shoestring budget vacations per year for mental health, etc.
I'd consider "being able to save enough to continue a modest quality of life for retirement" a normal feature of what describes middle class life.
If you are living paycheck-to-paycheck you'll suffer when you hit old age or have a disabling condition that prevents you from working.
Thanks for a detailed response. I'd definitely like some extra income as a grad student, but I cannot earn outside of the university because of my visa restrictions. For all intents and purposes, you may as well consider me living paycheck-to-paycheck. I wish academia paid better.
Why is this eating up NSF money / taxpayer money?! Megacaps like Google and Facebook are reaping tons of value from arXiv, i.e. (1) easy access to non-industry peer attention through a respected pre-print publisher, (2) major incentive to their employees and key component to performance review cycle, ...
Google itself has out-published most universities in conferences like NeurIPS for the past few years. For Google alone to match the $10m would be a drop in the bucket-- a $10m grant would be an order of magnitude less than the compensation they're paying to their own employees who make heavy use of arXiv.
NSF funds are scarce, precious, and it's immensely hard for lawmakers to divert funding to NSF versus e.g. defense. One thing arXiv needs is high-quality bandwidth and Google / Facebook have tons of that. Taxpayers are already paying tons for Google's several anti-trust trials, no need waste NSF budget for something industry can very very easily solve.
Seems a reasonable use of NSF money to me, to promote sciences by creating a common infrastructure for organizing the world's library of preprints. It's infrastructure of a sorts, and civil government is the ideal party to fund civil infrastructure. (Isn't it?) Not random FAANG's, and I don't see why they'd want to anyway—what they'd get out of it.
FAANG's aren't charities; anything they do that looks philanthropic is some kind of trap.
I’m not saying it isn’t a bad use of NSF funds, I’m saying that especially in the AI space that industry use of arXiv is large enough that taxpayers deserve to have industry foot the bill.
I favor the "company pays tax, tax funds science infrastructure", cycle, not the "company funds science infrastructure" cycle. It reduces the risk of industry asserting influence over the infrastructure.
Whether companies pay enough tax is a separate discussion, but they paid some taxes, and those taxes are now funding ArXiv. This seems like a good setup.
But I think you have it backwards: public infrastructure exists to make it easy for industry to thrive, to create low-friction business environments. We like industry, and we want more of it. The US benefits from AI companies headquartering in its borders far more than the AI companies benefit from "freeloading" off of arxiv.org server costs.
And let's not lose perspective: the stuff posted on arxiv.org is collectively like a million times more valuable than the servers themselves.
For me personally, it’ll be really interesting to see if the frontend changes. No doubt there are some important improvements that can be made (a website can always be made more accessible, moderation tools, support for name changes as mentioned, etc.) However, to first approximation, arXiV’s website already seems almost like a platonic ideal. It reminds me of Craigslist. Simple HTML, loads fast, has the information and features you need but otherwise gets out of your way. I love it.
The arXiv team deserves a lot of credit for what they’ve done to get it this far. It’s difficult to overstate how useful and transformative preprint servers have been to science.