Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
Let’s start a new thread to talk about this. One place to start is brainstorming what-all we could do with grant money. Here are some ideas:
Other ideas?
Personally, I have very little experience writing grants, and zero experience starting with an idea for a proposal and seeking out a potential funder. But I’m happy to give it a go if others are seriously interested, and I can get help from the grants office at USD.
We could have a workshop (possibly partly virtual) where a particular cloud of topics in the nLab is worked on collaboratively to include proofs, updated (or more thorough) references and so on. I’m thinking of something like the Stacks Project workshop, or a Wikipedia Meetup.
First, what’s the perspective regarding our arrangement with CMU? Do you know?
How much longer will it run? Are there any options to continue it further?
Urs #3, I think I answered that question in another thread: the server was purchased with grant funds, so it will remain around until the physical machine dies, but not after.
I like the ideas in #1 and #2! I imagine that we would have a better chance if someone with a permanent position within academia is the main contact person, which I suppose means Urs, Mike, David C, and Zoran amongst the regulars. It is indispensable that Urs is fully involved, and that Zoran and David C are invited to contribute as fully as they have time for, but my suggestion would be that Mike, who has experience from the HoTT book of coordinating collaborative work, might be asked to ’lead’ the application if Urs supports this. By ’lead’ I just mean take the responsibility for organising things, be the external point of contact, etc.
It is of course also crucial that all regulars have a chance to give input. I am happy to contribute what I can. I would suggest just to get cracking on trying to write down some kind of application as soon as possible; maybe we can use git or something (as well as email for those not comfortable with git). Tim especially has lots of experience from grant applications; it will be very good to ask his advice and ask him to contribute.
I had a quick look at the latest call for applications from the Norwegian research council yesterday, but didn’t find anything that fitted. I don’t know how things work in the US; maybe someone who does could take a look? There are also places such as the Simons Foundation, but I am slightly uneasy about these kind of private benefactors; but if others are fine with it, I will not raise objections.
Regarding students, this sounds great. Often institutions just have a certain amount of money for funding summer jobs, and one can propose more or less anything as a specific job, which will then be advertised; is there any chance that this is the case for your university, Mike?
As for potential funders, it would make sense to think in terms of primary beneficiaries of the nLab. In terms of disciplines the order might be: mathematics, (mathematical) physics, computer science, …, philosophy,…
As for my subject, I doubt a Humanities research council would take it up. Surely it’s the start of my list to think about first, either in terms of government bodies or private bodies, such as FQXi. An alternative might be something like the John Templeton Foundation, which funds big picture research across disciplines. For the latter, Todd’s description of the nLab as like a mediaeval cathedral (here and later here) might go down well.
I hope you are all aware that applying for grants is a heavy form of gambling. Do we have any indications of any grant agency that they might have any inclination to fund something like the nLab?
It would be good for our health and sanity if we had a concrete idea of some person supportive of the idea, a person with some power or even with actual money in their hands. Like Steve has been.
Hi Urs, I only have familiarity with the Norwegian system, but if the right ’call’ comes up, I would expect there would be no problem at all for the nLab to receive funding from the Norwegian Research Council (I am not proposing this as the best entity, just as an example). Relatively speaking, we would not actually be asking for very much money. I believe something like $5000 was used to purchase the existing nLab server, which is trivial in relation to a typical grant size. So whether we can obtain a grant to cover all of what we could dream of is unclear, but to obtain a grant to cover something is very realistic in my opinion/experience. I would not describe it as gambling :-): that success in any particular application is often at the whim of whoever is on the committee is certainly true, but I see it just as a process that one goes through. It is obviously very frustrating if one does not receive funding when one feels one has earned it, but if we just accept that and keep trying, we don’t have much to lose as I see it, so long as everyone chips in and there is not too much of a burden on any one person?
It is of course fantastic that Steve has been supportive, he has done a great service to the mathematical community. But I think the nLab is perfectly able to be funded on its own feet (in fact, I think it may be easier to find funds for it than for funding a research project), and I think this would be healthiest in the long run both for the nLab and for Steve and others.
I agree with David’s thoughts in #6 by the way. Something to keep in mind is that I think one cannot just send an email to someone at the Simons Foundation and say “we’d like funding”! We would, I think, need to build a good proposal, and approach it quite formally. But it might be good with this kind of foundation to get in touch directly rather than use an existing funding programme/mechanism, because our case is a little unusual.
The following kind of funding programme might for example cover the nLab (see the third bullet point under 1.): esfri. Again, not proposing this itself, just an example.
Could you give us an idea of what we might want to be requesting?
I don’t claim to be an expert, but I would imagine an application would go something like the following.
Indicate the scale, maturity, scope, and influence of the nLab (e.g. indicate how many hits it has per day/month/year; it has a hit almost every second, so this will be high). Indicate its current funding model.
(Optional, but maybe a good idea) Get some testimony as to how useful the nLab is (maybe we need some kind of thing where we ask people using it to leave their name and affiliation and any comments to indicate that they find it useful; hopefully many people will contribute).
Indicate the benefits for the research communities relevant to us of the continued existence and further development of the nLab.
Indicate our needs. At minimum, we need enough funds to cover the running costs of putting the nLab in the cloud. We provide some concrete numbers here, and concrete indication of when and how the money will be used. We ask for support for this for n years, where n is appropriate for the type of grant we are applying for. We indicate the benefits of conferences/workshops, etc, and make budget estimates and concrete plans for how and when to hold one. (But workshops/conferences may be able to funded separately from the infrastructure costs; again, it depends on the type of grant programme).
If needed, indicate institutions willing to co-operate with us (e.g. CMU).
Thanks! Just roughly, what do you think we need or want to request per year?
It’s a bit difficult to say without looking closely into it. There are some prices here for a server hosted at AWS. A really powerful server (much more than we would need, I expect), say m4.10xlarge, is about $20000 for three years of use. We can probably make do with much less; we’d have to try out a few different servers first and experiment with the performance. We could also for example put the MySQL database on a different server; there are lots of possibilities.
So, purely for infrastructure (i.e. excluding workshops/conferences, etc), perhaps something in the region of $10000 per year would be a first guess (though we could no doubt get by on much less)? As I say, though, this is very tentative.
It seems that the German Volkswagenstiftung has funding possibilities for “unusual projects” including infrastructure in mathematics for German scientists. Since applications are only on invitation one is supposed to contact the officials in charge of the math grants per mail/phone making at least the first step of an application resp. rejection sufficiently low level and casual for an energetic German researcher related to the nLab.
Hi Richard, thanks for the information! Okay, so we are looking for somebody willing to spend 10K on the Lab per year. That’s good to know.
Thomas, I agree as in #7, that the right way to go about these things is to try to first find a person in power that is supportive of the idea. Hopefully there is that energetic German researcher related to the Lab somewhere!
(A while back, in another life, after I had compiled and submitted a full application for an FL- Humboldt grant (an inch-thick stack of paper when submitted, they really asked for paper mail, that should have alerted me…) I first heard nothing, then on request I eventually learned that they decided that I don’t count as a German researcher anymore.)
Guys, just to drop a note of support. I have some experience with proposal writing within various teams and am interested in doing my part of work in any proposals in future. I am now following this discussion on the side, having a pile of deadlines in next few weeks, especially right now preparation for some public lectures (and exams and so on) this very week so I will join this discussion later. Croatia also used to have 2-year infrastructural projects in science with informatics component which were relatively easy to get but then few years ago this grant program was abandoned. I will watch if they ever resume the program.
$10K per year seems to me like way more than we ought to have to pay for a cloud server. At a nonprofit where I volunteered as a sysadmin for a while, we paid more like $2K per year for cloud hosting, and that was with several distinct servers that we didn’t actually need but which were helpful for development. Perhaps the nLab would need a bit more power, but it’s hard for me to imagine we’d need all that much more — there aren’t that many mathematicians in the world. Do we have statistics yet on how much the nlab is actually used? How many page loads per day/hour/minute?
Another possibility would be to continue piggybacking on some other grant, as we did on the HoTT MURI. For instance, I might apply for an NSF CAREER grant next year (if I ever get my act together), and I could potentially include some nLab funding in the budget for that. Any others of us who apply for personal grants could try the same.
I’m hopeful that we could make a good case for funding the nLab on its own too, though, especially if we can accumulate some supportive comments from users.
$10K per year seems to me like way more than we ought to have to pay for a cloud server.
Yes, do not attach much significance to this number. My experience in a previous job was with a much larger collection of servers, with orders of magnitude higher costs. I suspect that t2.large might be enough for us, which is less than $1000 for three years. My reason for estimating more was that costs can quickly escalate in the cloud, and we do not wish to reach a situation where we have run out of funds or cannot scale. Also, there are many different models; currently we have everything on one server, but one could split this up; one could have several ’instances’ with a load balancer in front; one might have the database on its own instance; an S3 bucket for logs; the nForum on its own server; etc, etc. All of these things have a cost. So it depends on how robust we would like the setup to be, how much time we have, etc. At a completely different end of the scale, one could even fund maintenance of the nLab (not for me, but somebody could be found).
For instance, I might apply for an NSF CAREER grant next year (if I ever get my act together), and I could potentially include some nLab funding in the budget for that. Any others of us who apply for personal grants could try the same.
This sounds very good. You could have some time dedicated to working on the nLab, maybe you can fund a conference, etc. As you hint, I don’t think this should be a substitute for funding dedicated to the infrastructure, but it could be a very nice supplement, if it is something you or anybody else is interested in and willing to dedicate time to.
Also the summer jobs possibility is, I feel (see #5), worth investigating regardless of everything else.
Do we have statistics yet on how much the nlab is actually used? How many page loads per day/hour/minute?
Not properly. As discussed before, it is definitely something we need; I’ll add this as well to the TODO list.
From one of the logs for the recently added APIs, I can see how many hits there are purely for actual nlab pages (under /show), i.e. not revisions or /author or /cite pages, etc. In the last day or so, there were 14710 such hits. Its pretty continuous over the 24 hours. So on average about 10 per minute, which looks about right.
Google Cloud Platform has grants for storage and computation.
Thanks very much, Bas! I had no idea of this, and I see absolutely no reason not to apply for it (unless someone objects to Google being the source of funds) for the infrastructure costs. We could apply for workshop/conference funding through more conventional academic channels. Of course, once one has been given one grant, it is typically easier to obtain further ones!
Edit: depending on what they offer, it would probably be too risky to move the entire nLab over (it’s also not clear to me whether the grant covers a running server). But I’m sure we could make good use of the money.
Thanks for pointing that out, Bas. However, it looks to me as though it’s intended for projects of finite length that require computational resources, not for simply running servers with academic value.
I’d donate a few k if needed, just write me with wire data.
I had a similar thought: If we just collected a few K dollars from our pockets and bought Richard a server, to be set up in his apartment, plus whatever it takes. Then in 10 years, when the machines dies, we do that again..
It might indeed be less risky and stressful if we were to run our own servers, rather than relying on continual grant support for running in the cloud. The main issue currently is with robustness: with only one server, in a location not specifically designed for running a server, we are always at the risk of downtime or worse. If, however, we were able, through donations or whatever, to purchase a few servers to be placed say in differing universities across the world, and then set up some kind of load balancer in front, we should be able to achieve a good, robust setup, so that if one server goes down, or even several servers go down, no user notices. A minor additional benefit to having servers in differing locations (US, Europe, Asia, …) would be reduced latency, i.e. quicker loading times, for people who are currently far from the US.
We could buy one server for you plus one for, say, Mike(?)!
Having the servers with private persons should not be less robust than having them “in universities”. On the contrary, it would mean one layer less that could break. One sees this vididly in our present situation, where all our urgent inquires regarding our physical server run via email through a person who, while being very kindly helpful, couldn’t care less about the nLab.
Hehe, I’m happy to help, but my thinking was that in a university the server could be put in a proper server room, which hopefully has good procedures for handling things like power cuts (and also regulates temperature, etc). Maybe even backups (NTNU used to do this when the server was there).
On a purely personal level, I do not think that it would be practical for me to have a server in my actual home, as it is not too large, and I have a young, happy, and energetic son running around! (Also I think the internet connection is probably not fast enough, etc). But I’d be happy to try to coordinate putting one somewhere, probably either NTNU or where I work. It is probably good anyway to have other people with access, just in case something happens to me or anybody else who has responsibility for a server.
Yes, it’s likewise not feasible for me to have a server in my home. I would probably not even want it in my office. But I could, I guess, look into whether USD would be willing to keep it in a server room.
I hope it doesn’t make me sound unsupportive of the nLab, though, when I say that I wouldn’t want to donate money from my own pocket, nor do I think any of us ought to have to do so.
Maybe there could be a general appeal for donations, like Wikipedia often has? We could proceed with grant applications anyway, of course, as the more servers the better.
I was just about to suggest Richard’s #31. This might require first setting up some legal infrastructure like a 501(c) [a non-profit], but it could be done.
(Tongue in cheek, but not completely) What about an nLab Foundation? ( I was a bit involved with the initial stages of the K-theory Foundation, and I would certainly not suggest making things that complicated!)
I hope it doesn’t make me sound unsupportive of the nLab, though, when I say that I wouldn’t want to donate money from my own pocket, nor do I think any of us ought to have to do so.
On the contrary: The idea here is that the burden of dealing with grant proposals is likely much heavier than that of paying for just what we need.
By the way, to put things in perspective:
I did pay the nLab server out of my pocket for the first years, after I had initially set up the nLab. Then I had paid for it jointly with Andrew Stacey for a few more years. It is only since rather recently that we lucked out to have Steve Awodey branch off money from his grant for the server.
If the load on nLab servers is as described in #21 (1 page query per 6 seconds), then a simple cheap VPS will be more than sufficient. For instance, RamNode offers a VPS for $5 a month: https://ramnode.com/vps.php
I guess different people have a different idea of burdens. I would rather write a grant proposal than pay several thousand dollars out of my own pocket. (For one thing, my pocket also has family obligations to meet.) But I will not object if others want to foot the bill themselves.
Setting up a 501(c), on the other hand, is not, I think, a burden I would be willing to be involved with.
If the load on nLab servers is as described in #21 (1 page query per 6 seconds)
This was on average for one day that I looked at, there can be several queries in a second at times. Also of course the server is not just serving web pages, there is the Instiki and nForum applications, and nginx, running, as well as a database which has more load (several queries are involved when a page is loaded, etc). I do not know yet how to put good numbers on the load. It may be that the kind of server that you suggest will be fine, I do not know.
Of course we do already have a server which is working fine except when there is a power cut at CMU. So we are free to try out things with regard to introducing further servers and a load balancer.
There seem to be lots of different opinions! Not sure how best to proceed; I do not have strong opinions myself, I will try to help out with whatever is chosen. I do think we should try to do something, though.
To take note, I asked around and these were some hosting platform suggestions:
And the one already mentioned:
Hello all, haven’t been very active on the nLab lately but noticed this discussion and wanted to jump in with one thought: perhaps the OEIS could serve as a useful model? I’m not sure about their relative scales (maybe the OEIS is bigger by an order of magnitude or so??), but I think the nLab plays a somewhat analogous role across a broad range of communities. The OEIS is supported by the OEIS Foundation (apropos Tim’s comment #33), which was set up in 2009. The website of the foundation contains a lot of information which may be of interest, such as treasury reports. For example, according to the 2017 report, the foundation’s total revenue was about 53K USD, with a little under half coming from grants and a little over half coming from donations, while their total expenditures were about 20K USD.
In terms of measuring impact, the community also maintains a list of works citing the OEIS, though I am not sure how this is kept up-to-date.
May I suggest that we try to launch a copy of the nLab on one of those servers (e.g., ramnode) and see how it works out? This may be the easiest way to determine if the resources are sufficient or not.
@Dmitri - -But a “copy” of the nLab wouldn’t be getting the actual hits that the real one does, so it wouldn’t tell us anything about the resources needed. Unless you’re also suggesting to temporarily redirect ncatlab.org to the copy… in which case there would be rather the headache of syncing the databases before and after the temporary switch. Or am I misunderstanding?
Re #42: Synchronization in this case means copying the database from one server to another before switching, then copying it back before switching back. I presume this is not too difficult, but perhaps Richard Williamson can say more.
I am sure we could manage it one way or the other. If we expose the database externally on the current server, we can just call it from the other server.
Adeel handles the DNS stuff, so if people wish to try this we could discuss with him when it would be convenient.
There is a quite a bit of prepatory work to do though; first I would like to be able to build Instiki locally, and be able to ’deploy’ properly.
Also, I think I would be most interested in testing a setup where we have a load balancer and both servers are in use, as I think this is in the long run what we will be aiming for. The nice thing about the cloud, of course, is that it would be simple to add more servers as we need them, once we have a setup that works for two.
I suppose if we were willing to simply take the whole nLab down for the time it takes to copy the database and migrate the DNS from one server to the other. It would be better if we could avoid that though.
Calling back and forth between (widely geographically separated) servers for every database query seems to me like it would slow things down quite a bit.
I am now writing my CAREER grant proposal. Shall I include a few K$ for an nLab server in the budget? Selfishly, I would kind of like to do so, as I think the nLab (and associated things, like getting students to work on it) would be a nice “educational initiative” to complement the rest of the proposal. And without lab equipment and so on to buy, it’s hard enough for a mathematician to find enough things to spend money on to get up to the minimum proposal budget. However, I of course don’t want to do this without the support of the community; what does everyone think?
Another idea I just had is that I could include funds to hire a USD CS student to work on some of the Technical TODO List (nlabmeta) items.
Sounds good to me!
Both of these sound good! Also, funds for distributed servers would be nice, so that the lab doesn’t go down whenever the builders at CMU feel like it.
Very good! As I think I wrote before, I think a student is a very good way to get work done on the technical issues.
Yes, it all sounds very good to me.
Regarding your quest for items to extend your budget, I’m reminded of the joke:
Dean, to the physics department. “Why do I always have to give you guys so much money, for laboratories and expensive equipment and stuff. Why couldn’t you be like the math. department - all they need is money for pencils, paper and waste-paper baskets. Or even better, like the philosophy department. All they need are pencils and paper.”
So you got involved in math as a ploy to get a waste-paper basket, David?
Strangely enough, I’ve just been down-graded there. I moved office yesterday and where my old office had two large bins, including one for recycling, now I just have one small desk “caddy”.
Perhaps someone else knew of that joke!
I think Mike’s idea sounds excellent.
1 to 54 of 54