Hacker News new | past | comments | ask | show | jobs | submit login
Should I Resign from My Full Professor Job to Work Fulltime on Cocalc? (sagemath.com)
250 points by occamschainsaw on April 17, 2019 | hide | past | favorite | 151 comments

This is a hard question to answer, but here are a few thoughts:

1. Resigning from a tenured position likely means taking a one-way street. The exception is a well-known scientist that can easily switch tenure to 2-3 other school today. In many fields competition for tenure is brutal and once you leave you may have some disadvantages in re-entry against a bunch of post-docs. "Only losers leave academia" is still a widely held mindset. All this does not mean not to resign; just be aware.

2. Do you still need money? The fact that the company is not ramen profitable in 7 years (including last 3 years of dedicated effort) means it may not be a source of income for a while. If you are not set for life, how long can you afford to not work? Will you be OK in industry (some professors thrive there some are completely miserable with a manager who tells them what to do and imposes deadlines)?

Unless you are set for life I would seriously consider your 50% option -- you get another year to fully dedicate to your startup without teaching a single day and still maintain tenure. In a year, re-evaluate. My 2c.

> 1. (OP here) I doubt I could easily get a tenured position somewhere else today, because I have not published much in the last few years, due to focusing on CoCalc. I also I have a 2-body problem (my wife is an academic), which would make getting another position in academia that much harder.

> 2. Do you still need money?

Yes. In fact, I didn't save enough as a college math professor to retire while living in Seattle (an expensive place)!

> The fact that the company is not ramen profitable in 7 years

The company is not profitable, but it is "ramen profitable" according to http://www.paulgraham.com/ramenprofitable.html. Without getting into details, the bank balance goes down a little each month since we pay hosting and for four fulltime employees (we have about 2 years of runway). However, we're close to profitable. If we instead had only three employees (or paid ourselves less), then we would be profitable. If we had just one employee (me), then I would be making a LOT more money than I made working at UW, though I would likely be much less happy (the other three people working on CoCalc are fantastic!).

I'm a proud UW CS graduate so this story is thrilling.

It was mentioned at times in your article that you used your network to seek advice on the next steps for the business.

Have you used the UW network to seek business advice from people not in academia and not in the math field?

If you are supporting four people, not all of whom are developers, that's an amazing place to be, imho. If you are supporting four developers, it's even more admirable AND probably time to involve someone who can think exclusively about how to grow the business and does not think about engineering tasks. There is a huge network of former UW people involved in the software industry in Seattle who would help if you are at that moment to consider it.

Maybe you've already done this, your article makes it sound like you have been very thorough in investigating how to take this to the next level, but it wasn't mentioned explicitly.

I know some people like this, reach out to me if you want connections to them.

Here's to your future success!

Do not leave that tenured position!! The market has been pumped up so hard that you should be making stability-reinforcing life choices. There’s also just nothing comparable to the benefits of being a tenured professor in the modern labor market short of winning the lottery. Find a way to make it work with fewer people or look for other ways to increase your revenue/demand.

> There’s also just nothing comparable to the benefits of being a tenured professor in the modern labor market short of winning the lottery.

Having a few million dollars in the bank is comparable to being a tenured professor, at least in terms of guaranteed income. (Of course, I don't have millions of dollars in the bank.) Maybe that's what you meant by "winning the lottery", e.g., that phrase is sometimes used to refer to startup founders who sell their company.

No I mean the modern labor market is horrible outside of the very hard to get positions at the FANGs. Have you researched what working and work culture are like? Read some of the Rachel by the bay blogs. You have a quality of life as a tenured professor that you simply won’t anywhere else. Especially as a “startup founder”. There’s a reason the valley takes them straight out of college graduation.

Do Not Leave Your Professorship.

You will regret it.

This seems like a very warped view of the world. The US economy is doing pretty well. The main worry people have is that it is doing too well and will lead to a crash at some point. Unemployment is basically non-existent in many parts of the country. Wages are finally starting to go up, after being stagnant since 2008.

Putting up with poor working conditions is mostly poor decision making, especially in tech.

Inverted yield curve

Decreasing job creation numbers

Looming Brexit

Recent Fed decisions to raise rates and more recently leave them unchanged.

Investment choices, “beta”, should reflect a heightened risk aversion for people later in their careers. That’s just standard financial planning.

I could even make a case for stagnant wages suddenly increasing indicating irregularities necessitating caution in financial and career decisions.

Your startup seems to be doing pretty well (I would call it "profitable-able in a pinch") and you have a mini safety net with your wife's academic employment should provide health insurance indefinitely if you do take the jump. Also, with your experience you will also likely find job in industry if your startup folds, so you have fallback options, assuming you are OK with working for a corporation.

That said, I would personally still take your "one more year" option as it gives you an extra year to get your startup to profit without officially resigning. This seems to be a very valuable freebie. Did I misread your web post about this option or are you reluctant to take this for other reasons?

And finally, on your startup -- I would steer it to become profitable ASAP; for example if and when you are deciding between hiring a fifth employee (to grow faster) or getting profitable sooner I would take the second path. My 2c. And good luck whichever option you choose!

>The company is not profitable, but it is "ramen profitable" according to http://www.paulgraham.com/ramenprofitable.html. Without getting into details, the bank balance goes down a little each month since we pay hosting and for four fulltime employees (we have about 2 years of runway). However, we're close to profitable.

Are you drawing a salary out of this?

> Are you drawing a salary out of this?

Yes, though it is currently significantly less than UW paid me.

As a former tenured professor who recently left their position, I wish you the best of luck. It's a difficult decision, especially these days. My situation was different in a lot of ways, but overlapped in others.

I don't think there's a right or wrong decision in situations such as yours.

Presumably revenue is trending upward? If so, when do you expect your costs to be covered (such that the bank balance stops going down each month)?

Are you spending anything on marketing?

> The fact that the company is not ramen profitable in 7 years (including last 3 years of dedicated effort) means it may not be a source of income for a while.

This is always a flag for me for something that needs to be investigated. Personally I think the idea the guy took to stick with CoCalc was interesting but at the same time he should really really consider hiring a CFO, someone who really understands where the money comes in and how it goes out.

Such a person could tell him exactly how much "lifetime value" is needed from each customer in order to make the company a going concern. Sometimes that number is "impossible" in that it is way more than one believes they could get for the product. When that happens, you have to take a serious look at whether or not you're looking at a hobby or a business.

It is always difficult to see these sorts of discussions after they have been had, especially when there was information that could have been developed earlier on that would have informed the decision, but it is not uncommon.

From what I have gathered from mathematician friends, they are catching on to CoCalc. I for one think it has promise. Especially with the dominance of Python in CS department intro classes, a Sage-based math environment for math classes (e.g. abstract algebra / group theory) seems sensible. Once those programmes start adopting it, you are good to go.

But how much are they willing to pay?

(I'm the OP.) Quick answer: "about $15/student". Our pricing page is here https://cocalc.com/policies/pricing.html, and people and institutions do regularly sign up for all the subscriptions and packages listed there. They were all created based on demand from users. We get a handful of new courses each week, and few individual subscriptions each day, though volume depends a lot on the academic cycle. One of the reasons I'm confident to resign from my UW job is that the business aspects of CoCalc are definitely working.

I don’t see income figures for you from this. But a general rule is your salary is only a fraction of the total cost of your position to the organization after all the overhead costs are included. We’re talking 3x-7x your yearly salary. So ask yourself whether getting 7x your income is what you’re getting from this the year after you resign. And because you’re tenured, that needs to be a long term demand that won’t go away.

Edit-I don’t see any of the above in your product/write up. Further, I used the high end of the coverage range as professor positions usually have good benefits. Stuff is expensive individually.

I agree with hiring a CFO. It’ll cost more in the short term but could pay off in the medium to long term like you say.

That said, even if it fails the next stop would be FAANG and that's where talented programmers are that are impossible to find in academia.

A few years later, richer than academia could ever provide (I left academia and did not regret it)

I definitely don't currently intend to work at a large tech company. However, I live in Seattle (where there are a few), and most of the Ph.D. students I've had over the years do work at such companies, so I at least visit them regularly. Indeed, basically all people you see when you go to lunch there are way younger than me (I'm 45 years old).

Since CoCalc doesn't have any VC investment, and does have a solid and diverse base of customers, it's basically designed so that it is unlikely to fail unless I do something really risky. An example of something very risky would be raising venture capital or getting acquired -- we won't do that, since we have explicitly stated we won't to our customers and on our website, because some customers have been reluctant to consider using cocalc if we do take VC. In the worst case, CoCalc would be what some people call a "zombie startup" or "lifestyle business"... i.e., not growing much, and just marching on. However, I love working on CoCalc, and our customers love using it, so even that situation wouldn't be failure...

I'm 41 and work at Google, and I don't think age is a problem at all. As long as you don't expect to get better treatment because of your age, promotions will and salary will go to the best person not the older person.

FAANG in general are also pretty good about work-life balance which is good for us older people :-)

He is probably too old for a faang.

This is totally not true. I'm 41, work at Google and doing pretty OK. As long as you are willing to keep learning and can use your experience constructively I don't think age is a problem here at all.

I was 62 when I worked at Google. I never felt any age discrimination there. It was a nice experience, not my best job ever, but nice.

> He is probably too old for a faang.

[many expletives later]

Really? REALLY?

Seven of 18 largest tech companies have a median employee age of 30 or younger.

If the comment had been about discriminatory hiring practices, or even just generally about ageism in the industry, I would have nodded, and probably upvoted.

But he phrased it as the OP being "too old for a FAANG", which centers the fault on the OP.

Besides which, in whichever sense it was meant, the comment was almost certainly factually incorrect. A math PhD with significant software engineering and engineering management experience isn't someone that gets turned away that easily.

I'm the OP, and I do wonder if I would be too old for a FAANG, with the fault being on me. Building a company as an entrepreneur at age 45 (my age) feels very reasonable, and I've seen research that suggests that my age is perhaps even a sweet spot for building a company. In contrast, I think I would find it difficult to work as an employee at a FAANG company, since it's very different than everything I have done so far in life, and most other employees would be younger than me (informal observation based on having lunch at Amazon, Microsoft, Google and Facebook regularly).

More importantly, joining a FAANG company would mean completely giving up on my longterm goal to massively fund Sage development, and that would be very hard for me to do.

Please don't buy into the industry's ageism.

While most other employees at any tech company would be younger than you, is having lunch at AmaFaceGoogSoft really so different from having lunch in a university campus cafeteria?

I am not trying to change your decision, BTW. Your long-term goal is an admirable one and I think achievable, whether CoCalc remains a private company, goes public, or is acquired.

According to the Stack Overflow survey, the median age of a professional developer is 30.

The tech industry has been growing rapidly, so a low median age isn't evidence of age discrimination (I'm not claiming there isn't any - just pointing out that a low median age is what you'd expect to see even in the absence of discrimination).


Which means that 11 of 18 do not.

The average age at google is 41.

But the average age hired is lower than 30. The whiteboard /stanford/group hire filters helps ensure only the best past a certain age has a chance.

"Only the best past a certain age"?

Google's own research shows no correlation between "elite school" or "grade average" filters and job performance. Their hiring comittees (from which the hiring manager is excluded) also make decisions while blinded to factors such as age, race, and gender.

Thank you for pointing this out.

> In many fields competition for tenure is brutal

Why so? There are so many people who would like to study but can't get into an university. There (as you say) are so many professors who would like to teach but also get into brutal competition. Why not just build more universities? Also why not start a private education company that probably wouldn't be eligible to issue government-accredited degrees (who but medical doctors and lawyers needs them anyway?) yet would provide university-level education quality and compete the free-market way?

The barriers (at least in the US) to university study are seldom admission. There already exist many accredited institutions that admit essentially all applicants with a high school degree or equivalent. The things that stop people who would like to get a college degree are more often money and time. Incidentally, these are the same things keeping would-be professors from tenured positions. Someone has to pay the professors, after all.

I think OP was getting at "valued" degrees, eg: A degree from a name brand institution, where a good chunk of employers will recognize the name of said university.

Degrees from non-name brand institutions (eg: all the community colleges/2nd tier colleges like Western Washington University) are for checking off the requirement of having a degree. Its a serious put down, but the name brands thrive on being "selective" yet bribe-able.

There is quite a bit more value to be had at the latter group of institutions, tuition is ~$500 a class or ~$1500 a quarter at community colleges and of a similar caliber in quality, much more reasonable than the $1500+ charged per 5 credits at name brand colleges.

In the US it is very easy to get into a university. It is not so easy to get into a good university.

> Why not just build more universities?

Because enrollment and the numbers of students who can pay are dwindling [1].

> Also why not start a private education company that probably wouldn't be eligible to issue government-accredited degrees (who but medical doctors and lawyers needs them anyway?)

The main reason anyone in the US goes to college these days is to get a degree, not to receive education related to any specific job. They want to get a degree because most employers still require a degree either for consideration or for promotion. The employers require this because they see it (fairly or unfairly) as an easy way to weed out those who would be unqualified. Any employer that would accept a non-accredited degree from a private education company that could easily just be selling certificates (which already happens anyway) probably wouldn't even care if you had a degree in the first place.

> yet would provide university-level education quality and compete the free-market way

I just don't see there being a significant market of people willing to pay full tuition for a non-accredited institution, to the extent that it could afford to hire good professors away from existing universities and be a better option for them than pursuing tenured positions elsewhere.

Compounded with the unlikelihood that students would be able to get reasonable loans to attend a school that could very well be a scam, this seems completely untenable. Students get into enough debt as it is going to accredited institutions with government loans.

1. https://www.wsj.com/articles/americas-disappearing-private-c...

> In the US it is very easy to get into a university. It is not so easy to get into a good university.

Because the academic job market is so tight, many of the professors at obscure poorly ranked universities are amazing teachers who do solid (but perhaps not famous or well promoted) research.

You should probably replace “good” in your sentence with “prestigious” or the like.

> many of the professors at obscure unrecognized universities are amazing teachers who do solid (but perhaps not famous or well promoted) research.

If I were choosing who is going to teach me I wouldn't really care about what research do they do, I would care about how good they are at teaching.

If you are the university dean or administrator who hires the professors, you maybe prefer the hot-shot professor who brings in a boatload of research grant money over the professor who gets the best teacher ratings.

The university gets a cut of the grant money and gains a bunch of reputation points for putting out cutting edge research. High teacher ratings are great, but the big universities care less about these and more about the number of Science and Nature papers a professor produces.

Unfortunately students aren't involved in hiring decisions. Everything I've ever heard suggests that faculty at prestigious universities are universally there thanks to their research credentials, not their teaching credentials. This was certainly true of the school I went to.

You know that unaccredited private universities (besides the phoney diploma mills) do exist, right?[0] The discussion shouldn't be about whether they can exist, but rather how much recognition they can garner across how broad of a spectrum of disciplines -- which goes up and down over time. I think we (as a civilization) are about to lose faith in academia, and I expect more opportunities in this space to open up, but I could be totally wrong.

[0]: https://en.wikipedia.org/wiki/List_of_unaccredited_instituti...

Edit: s/rime/time

> Because enrollment and the numbers of students who can pay are dwindling > I just don't see there being a significant market of people willing to pay full tuition for a non-accredited institution

Aren't US tuition fees already bloated?

> afford to hire good professors away from existing universities

Why not hire the professors that are still good yet not perfect enough to get through the brutal competition mentioned above?

> The employers require this because they see it (fairly or unfairly) as an easy way to weed out those who would be unqualified.

Why is it not possible for a private non-accredited college to establish an industry-recognized degree of reputation?

> Why is it not possible for a private non-accredited college to establish an industry-recognized degree of reputation?

You're absolutley right! This is very possible.

In practice, the process you describe is both more difficult than accreditation (because it's undefined) and less trusted by industry (because there's nobody trusted validating it).

Not always. Think programming language certification. If the company who created x offers certificate in x trust can be there. An exam is dufferent to a degree but it could work if say Apple or Microsoft started to offer a university and focus on tech like ai, etc. New York times offering an English degree. Scotland yard - security.

>Why is it not possible for a private non-accredited college to establish an industry-recognized degree of reputation?

It is absolutely possible. In fact, it doesn't even have to be a private institution. I have a CS degree from a public California university. The university itself is accredited, but my particular degree is not. There is no outside body that says that I learned anything at all about CS, just that I learned the bare minimum to get a bachelors degree from a UC school. The weight of the universities name is more than sufficient on its own to answer any question of whether my degree is worth anything.

Of course I suspect that it is in fact much easier to go down this path if you're already a globally recognized research university than if you're "Joe Schmoe's Computer Science Degree Emporium".

> Why not just build more universities?

Because universities are fortresses designed to protect the weirdos who can't survive in the real world. We're passing through this awesome enlightenment phase where there's room for everyone. But you can't just throw one up on a thursday afternoon. Universities need a kind of social contract. (i realize the one true scotsman weakness here, but i think the point holds)

Universities are where you stick the rich kid that's not strong enough to run the family business. Sometimes, you don't even have to be rich.

The thing that's neat is sometimes an idea pops out of a university that's pretty good. Newtonian physics and evolution spring to mind. i'm obviously nuts.

i would encourage you to read [1]. I'm not exactly making an appeal to authority, it's more of a here's a much better, less cynical interpretation of what i'm trying to say.

[1] https://www.cs.utexas.edu/~vl/notes/dijkstra.html

Universities in the US are no longer institutions of higher learning; they are hedge funds with affiliated football and basketball teams. Because of this, universities have decided to drastically cut costs by replacing tenure-track positions with adjuncts. About 58% of faculty at American universities are non-tenure track, and that number is likely to increase over the coming decade as tenured professors retire [1].

[1] https://www.aaup.org/report/status-non-tenure-track-faculty

Not many companies are going to hire an engineer from a non-accredited university. Sometimes it is specifically called out in the job posting. Accreditation is there for a reason, otherwise you end up with a bunch of people with degrees from diploma mills and ITT-like organizations applying for jobs they have no business being in.

Perhaps but a particular private company can establish reputation that would ensure whoever graduates from it actually has acquired the knowledge and skills listed.

In fact I even believe that teaching and skill/knowledge-certifying companies should be separated entirely. A person should be able to acquire knowledge whatever way they find comfortable (e.g. choose a company that would teach them or study entirely on themselves if they are sufficiently resourceful) then go to a reputable institution that would test them and issue a diploma.

Who is going to want to attend a school with zero reputation

When its job is to teach rather than to issue a diploma which other parties would recognize, attending a couple of lectures would be enough for me to see how well do they teach if it's worth my money. There should also be money-back warranty for those unsatisfied with the quality of teaching. At some point, if I were satisfied, I would also recommend it to people I know and such a recommendation would be enough for me to give it a try if I were them.

Measuring such a subject by thyself may seem like a fallacy as many people obviously don't think like I do but I believe there are many (a number sufficient to sustain a business) who do think the same way in such a big country.

Government jobs in particular often require degree (not university) accreditation for real engineering degrees. They generally don't require the same for CS though, so there are plenty of tech jobs that don't need accreditation at all. I'm a government contractor working as a software engineer with a non-accredited CS degree from a UC school. If I had a degree in engineering, it would absolutely need to be accredited (and my university's engineering degrees are).

Some government jobs do want the accreditation--you need it to become a patent agent, for example--though others don't. I think it might be related to whether it's officially an "engineering" or "science" job.

It's a little aggravating though, since many good programs (e.g., Stanford) don't bother with the ABET stuff for computer science.

I mean many companies don't care if you have a degree at all, let alone where that degree comes from.

> My third year of unpaid leave from UW is up. I have to decide whether to return to UW or resign. If I return, it turns out that I would have to have at least a 50% appointment. I currently have 50% of one year of teaching in “credits”, which means I wouldn’t be required to teach for the first year I go back as a 50% appointment. Moreover, the current department chair (John Palmieri) understands and appreciates Sage – he is among the top 10 all time contributors to the source code of Sage!

> I have decided to resign.


To be perfectly honest, it's not that risky. It won't be hard for him to get an academic job somewhere if it doesn't work out. This is usually not the case, but for someone with his background, finding another academic job won't be difficult if he's willing to work outside of a highly-ranked research university math department (and the probability of being able to get one of those jobs is even good).

Well, they got to find a position for the wife too; prepare for trouble and make it double.

> I have decided to resign. I’m worried about issues of intellectual property; it would be extremely unfair to my employees, investors and customers if I took a 50% UW position, and then later got sued by UW as a result. Having a 50% paid appointment at UW subjects one to a lot of legal jeopardy, which is precisely why I have been on 100% unpaid leave for the last three years.

This is a really easily solved legal issue. You draft up a quick little IP agreement that clarifies the desired state of affairs - problem solved. Resignation was not required, needed, and arguably made the legal risk worse.

Oh well.

UW C4C would likely get involved with such an agreement and pitch a fit, UW can be a vicious beast. They don't care about the community around them, beyond what will make UW $$$ or make UW seem more like an Ivy League. This recent dispute is illuminating: https://www.seattleweekly.com/news/workers-protest-uw-laundr...

Another thing I've seen go down was UWave.fm got a LP-FM license from the FCC, had a tower site and was lined up to set up a radio station at UW's Bothell campus, and the board decided to sit on the FCC approval till the very last day before the FCC needed hard dates on tower build and frequency use timelines, only to tell this student group that UW was canning the whole thing (despite it not costing UW a dime).

I've worked with plenty of patent licensing and tech transfer offices at universities. This isn't a situation where they'd pitch a fit. Worst case scenario, they need an email from a dept head directing them to get it dealt with.

Your average legal dept is overly conservative, not brain-dead.

I think OP thought navigating office politics to accomplish what you describe would be quite troublesome, and they would rather bank on the current department chair keeping C4C from taking action after they leave, rather than fight for basic rights inside UW.

UW is not kind to their employees: https://www.usatoday.com/story/college/2015/04/23/4000-uw-st...

I've spent a lot of time on exactly this sort of thing over the years, with very strong support from the math department. There are an enormous number of rules (laid out at Executive Orders like this: http://www.washington.edu/admin/rules/policies/PO/EO57.html), and the university works within the framework of all of those rules. So even if everybody agrees that it would be far make the most sense to do X, if the rules explicitly say "you can't do X", then you don't get to do X. It's very different than how things work in a small company (say), or maybe a smaller private university. UW is a massive (and old) state university, and isn't so flexible. For example, when I asked about being appointed at less than 50% time, we discovered here (https://ap.washington.edu/ahr/academic-titles-ranks/professo...) "Must be appointed at a minimum of 50%.", and that's that.

There are quite a few professors I've run into at Seattle Central and North Seattle College that have formerly taught at UW Seattle, but left due to situations like what you describe (dept is fully supportive of X happening, but due to an age old rule it will never happen).

UW has a ton of baggage and is quite inflexible compared to other institutions in the region and state, their lack of flexibility is enabled by having the highest tuition in the area ($11k vs $4k per year) and significantly higher state subsidies (UW, WSU, etc taking 60% of the state's college budget, yet only educating 40% of the students) than 2 year institutions.

This underfunding of 2 year institutions in favor of UW and ilk was what triggered the walkout yesterday and the ongoing protests: http://www.thestand.org/2019/04/join-aft-members-april-16-wa...

FWIW, I think the title for your piece is doing you no favors. Titles are hard, so that's not intended as some zinger. I think it just set up a lot of people to think "Must be some flake that he would ask the internet to decide this hugely important personal thing for him!"

I only skimmed the article to try to determine if you were, in fact, asking the internet to help you decide. I got to the announcement that you had decisively decided and went "Meh, could use a better title."

Best of luck in your future endeavors, from an internet stranger who had never heard of you before today.

You're right -- it's a bad title, and your constructive criticism is appreciated.

It wasn't actually intended as constructive criticism per se. Just observation concerning some of the replies I'm seeing posted on HN, which strike me as being influenced by the title rather than rooted in the obvious fact that you've had years to think about this and know what you are doing.

You assume they would sign it. That quick little agreement is likely to be a major point of contention.

UW has tried overall to be as supportive as they can be within the parameters of Washington State law and University rules. I did get UW to sign something a few years ago, but it was clear that whatever work I did on CoCalc had to be on my own time, not using any UW resources. There's much more to why I resigned than just my fear of potential IP issues (note: the issues are not worse by resigning, because I have been on 100% unpaid leave for the last three years, starting with a BSD-licensed codebase). An additional factor, which I didn't explain well, is that I would have to devote at least 50% time (so 20 hours a week) to work for UW, and this work would not be on CoCalc. Unfortunately, I have found that I am unable to do a good job working on CoCalc, if I have to spend 20 hours a week focused on other things. It's also impossible for me to respond to critical customer needs, server issues, etc., if I'm preparing to teach (or actually teaching) or in a committee meeting. For example, imagine that a 500-person course at UCLA has an exam coming up using CoCalc, and there is some issue with pushing out the exam to students, which I could easily fix if I could focus on it. I would feel terrible ignoring the UCLA course for a committee meeting or because I was teaching.

This makes a lot more sense, Prof. Stein. Best wishes in your continuing quest :).

The article is a list with many incorrect decisions (like going to commercialisation office before asking anyone who had tried, etc), so I guess it shouldn't come as a big surprise they keep making them?

He’s already made his choice now he has to understand it. (Matrix reloaded)

This is correct. I appreciate all the comments in this thread.

I absolutely think you made the right decision!

I look up to you because you are not just another random math professor producing papers that no one ever reads. Instead you take a risk and try to succeed at what you are most passionate about. In my opinion this is both more noble and more exciting than being a pure math professor. I sincerely hope CoCalc works out well for you, good luck!

I saw Prof. Stein's talk at JupyterCon last year and was blown away by what he had accomplished with CoCalc. It was one of the best talks at the conference IMO. Very excited to see where CoCalc is headed and wish Prof. Stein all the best.

Edit: here is the talk https://pyvideo.org/jupytercon-2018/real-time-collaboration-...

Thanks for the link. Watching it I can't help but feel a little sad. Prof. Stein is a highly trained mathematician with a track record in establishing successful software projects like Sage that many people benefit tremendously from (watching the video I learned that he actually started Cython as well!). It says something about the state of affairs -- the general lack of awareness and interest in the development of math unless it benefits some business directly -- when he could not find the resource needed to develop the math software he loved the way he wanted to. Maybe he really enjoyed working on the sync protocols, text formats, backend databases and file systems etc needed to get CoCalc running smoothly. But I wonder if he had the resource he needed would he make the same choice? I say this as I have a son who is double majoring in CS and math. As a freshman he is already mostly done with upper division math (3 A+ in honors real and complex analysis) and is taking graduate math classes. Yet it seems prudent for him to at least put in the minimum effort to get his CS degree to keep his option open. Eventually it will be a hard choice down the road for him as well.

> But I wonder if he had the resource he needed would he make the same choice?

If I had the resources I needed, I definitely would not have made the same choice. Though I enjoy somewhat working on sync protocols, databases, etc., it is only a means to an end. Mathematics software is much deeper and more interesting me. If I had sufficient resources a few years ago, I absolutely would have instead built a team to work fulltime on Sage, similar to how John Cannon successfully built an amazing team around Magma, partly leveraging that Magma is closed source and commercial -- https://magma.maths.usyd.edu.au/.

Thanks! I'm currently working on realtime collaboration for ipywidgets, which is an interesting challenge.

As lead developer on Axiom, an open source computer algebra system, I can fully appreciate his funding problem.

One problem is that open source software can't really handle grants. Government grants require accounting to handle receipts, and follow all of the rules. We need a couple "open source accounting" people who can handle managing grants (just like a school's bursar's office).

I have tried to find a funding stream for nearly 20 years, without success.

I am currently doing computer algebra research as a visiting scholar. That is an unpaid position. Perhaps you can remain connected to the department as a visiting scholar.

If there is some way I can help, please let me know.

Organizations like Open Collective[1] can help to handle this.

P.S. What do you think about moving the site to GitHub Pages instead[2] (and overall modernization of it).

[1] https://opencollective.com/

[2] https://github.com/daly/axiom/issues/13

I’m not sure if SPI has supported member projects with receiving government grants specifically, but they are setup to provide shared resources like accounting for open source projects.


> Perhaps you can remain connected to the department as a visiting scholar.

I will, though as an "adjunct". It's of course unpaid, but makes it technically possible to do some things with the university (e.g., be on a Ph.D. thesis committee).

I know this isn't the most appropriate place to ask, but what are the main feature gaps between commercial computer algebra systems like Mathematica/Maple and systems like Axiom/Sympy?

Several comments.

First, Axiom is a research platform (currently working on proving the algorithms correct). It is no longer a commercial competitor to Mathematica and Maple (as it was in the 90s).

Second, Axiom is based on Group Theory, giving it a reasonably sound mathematical superstructure. Mathematica is based on rewrite semantics. You can rewrite anything to anything. Maple is based on tree semantics. You can re-arrange a tree in any way. Obviously, their computational mathematicians are well-versed and clever enough to avoid most obvious failures. It is just easier in Axiom.

Third, Axiom is open source. Consider what happens to computational mathematics when the companies supporting closed source disappear. (Companies die, on average, after 15 years). Macsyma (Symbolics) is gone. Derive (Soft Warehouse) is gone. I am deeply concerned about the impact on computational mathematics when this happens. Axiom, as open source, won't disappear as there is no company behind it. It will, of course, probably stall once the key developers stop maintaining it.

Mathematica and Maple are amazing systems. If you need what they provide I strongly recommend buying a license. I know some of the developers from these systems and they are extremely clever people.

If you're looking for competitive software, I'd look at Sage and CoCalc. This is an open source alternative to Mathematica, Maple, Magma, etc. It is open source and (mostly) free. I can strongly recommend it.

As for "feature gaps", there are many. If what you want are "features", Axiom isn't for you. But if you want answers that are proven correct, stay tuned. Computational mathematics IS mathematics. You can rightly demand that the results are proven correct. That seems to me to be a required "feature".

Thanks for taking the time to reply.

I am 100% in support of software like this being open-source. Not just because companies go out of business and disappear, but for scientific-computing software, it's important to be able to extend these systems. When I talk about features, I often mean can I rewrite this integral into something more computationally tractable? Or is there some reparameterization of this function into something differentiable?

I know this are very much how someone might pose a problem coming from Mathematica or Maple, but really I hope it's clear the end goal is still the same. Discover an equivalent mathematical representation of my problem that's more amendable for further work.

You should copy Magma's model but be open source and use support or web hosting (like cocalc does) to justify the price tag. Then you get users to give you their grant money. Or convert to closed source main branch and release 1 yes old versions or at $ bounty milestones as open source, to motivate users to buy a license.

This post resonated with me strongly. Yes, resigning sounds like the obvious best choice. I hope it works out well!

I can certainly understand why Prof. Stein doesn't want to trash his colleagues, but I'll come out and say it: if academia wasn't completely dysfunctional, it would find a way to support this work.

For my own work (which is nowhere nearly as academically relevant as CoCalc and Sage) I'm planning on doing it myself without worrying too much about funding. I can use a combination of savings, random consulting gigs, Patreon, etc. I can imagine doing more "company" kinds of things, but only to the extent it supports the work. This is not an easy problem, and again I wish Prof. Stein well.

If Jim Simons wouldn’t play ball it’s hardly only academia that’s implicated

Jim Simons doesn't understand open source and open access, because the source of his $15Billion net worth (proprietary hedge fund strategy) depends on him not understanding it. And also perhaps because he doesn't understand software very well. He could have made Magma free for everyone, but instead he bought a license only for his chosen elites to use.


This is true. The philanthropy scene is probably one of the few areas that has even more screwed-up incentives than academia.

I'm really disappointed (with my impression) that UW doesn't seem to appreciate what has been accomplished with Sage and the value of a major open source competitor to Mathematica, Maple, Matlab, etc. I hope William and Cocalc are tremendously successful.

If I'm reading the article right, he's only done work for UW for one out of the last 5 years. That seems to me to be an enormous amount of flexibility on their part.

Yes, that's correct. The math department has been as flexible and supportive as they can possibly be. They won't let me be on unpaid leave again only because the university won't. However, it's not like the mathematics department at UW has a lot of money just laying around. I don't think they have even a single position that is paid for by an endowment. It's very different than other departments I've been at. I was Harvard University's math department for five years, and they had far more discretionary money than UW math; also other organizations on campus such as Computer Science have received significant donations. The mathematics department's money comes mainly from the school allocating funding in exchange for teaching many important courses.

Yes, but you'd think the university could find a way to directly support building the software.

I used to think that too. I only started CoCalc when I realized after a decade that this probably wasn't going to happen (for me).

UW as an institution cares mainly about getting all that Silicon Valley they saw Stanford get from Google and followers, so they can find administrators and build fancy buildings, not about research and education. Non-commercial research and teaching has been on a long decline, sustained only by previously tenured professors.

It's a university. Work for UW is supposed to be generating valuable research and education, which is what Sage and Cocalc do.

I think this is a sad reminder that as a developer just because you add value, doesnt mean you will be able to capture it. There is a tragedy of commons at play for tools, most of us (including companies) are unwilling to pay good money for good tools.

To add to his accomplishments, Sage was a pioneer of the web-based client/server notebook model now popular in Jupyter. I'm not sure if Sage was THE first instance of this, but certainly one of the first.

Well to be fair Mathematica had a client server notebook architecture since early on. Today it also supports a web fronted besides the now (QT based?) native client. The native client is way more powerful and performant than anything that Jupyter offers and can be extended in the client Language itself. IPython/Jupyter is a more or less direct copy of Mathematica's interaction pattern but as is often the case a poor imitation in many regards.

As a comparison https://observablehq.com seems like a genuine novel take on the notebook pattern and is unsurprisingly a commercial effort with some open source components.

Thanks! You're referring to the Sage notebook, and the key feature is that it is web-based with some collaboration support. I believe it was the first instance of this at the level of sophistication that it attained, which was significant given the web development tools available a decade ago. I started sagenb in 2006 after seeing a demo of a local OS X frontend for IPython. I did a huge amount of work on it, but there were also many students who did too (e.g., a 16-year old in the Philippines named Tim Dumol). The Sage notebook (and IPython) was very much inspired by Mathematica, Macsyma, and Emacs. Jupyter notebook was written a few years later, and has a more modern architecture (it used websockets). Jupyter has a much newer client called JupyterLab, which uses more modern Javascript tooling, including Typescript, WebPack, Phosphor, and so on. I've been working hard to make CoCalc fully support the Jupyter ecosystem, since I love the momentum behind it and how inclusive the community is. In particular, I wrote a new Jupyter client that supports realtime collaboration.

That's what happens when faculty give up administrative decision making power to a professional bureaucrat class.

Did they give it up, or did the Legislature hijack it? Who appoints the people who run the university?

Professor Stein is an active HN member by the way: https://news.ycombinator.com/user?id=williamstein

not only that, his actual post of this blog post is here https://news.ycombinator.com/item?id=19665027

I wish I could come up with a more helpful comment, but, as someone with 1-3 terminal windows dedicated to Sage pretty much at all times on my machine: no, giving up a tenured professorship for Cocalc sounds like a bad plan.

Yes, you are absolutely correct in your decision to resign.

It's - if I'm being honest here - painfully clear you are much more interested in your side project (from the prospective of UW) than your ostensible job.

I think for your sake and theirs' you should resign and try to commercialize this, or you will fail at both. Either that or completely hand off or abandon the project.

It's not inherently a side project, it only has to be one because his department doesn't view this work as a contribution to pure mathematics. Seems shortsighted for a field where progress is often measured in centuries.

Wow, I just went to the Cocalc website, and it's truly shocking. It seems like an interesting service, but terribly marketed. Do you have any designers? I know your service is targeted towards technically minded people, but it still needs to be able to market itself. Have you done any user testing on how people use your marketing website? Here is my experience so far:

The first thing I see is a NINE minute long video. As a first time visitor who doesn't know anything about this service, that's too much time to invest.

The second thing I see is that it supports a lot of stuff. Cool, but why is it better than what I have now? I see that I can share things privately. I can already do that. I can "time travel". Yes, version control is a thing. I can publish it online. Great, I already do that. These are the only thing I read because they are the only thing that's bold. To be honest, this is where I would have clicked away.

It also doesn't help that the navigation is unusable at half-laptop screen width, let alone on mobile.

Only after scouring your website do I realize that it's basically Google Docs but for math, which is cool! But this was not clear at all when first visiting your site.

If you're interested in more feedback feel free to reach out.

UW is a prestigious university and you already earned tenure. Even if you quit your job and the project fails, you're going to be just fine. Probably still a scary decision, but in my eyes you have already "made it".

(I'm a tenure-track first year prof.)

Is it really that easy to get tenure all over again, after a failed project/startup?

Many, many schools would hire him with tenure if he decided to go back to academia. Or he could choose to go to industry and make plenty of money.

Plenty of lucrative and fulfilling options.

Can you elaborate? Is this field-dependent? This is completely against my (very, very limited) experience:

I got my PhD ~20 years ago and saw a few tenured professors (math and physics) leave academia. While I have no first hand experience I did hear through a grapevine that some explored options for coming back, but could not. This is a sensitive topic with little data and so my info is certainly suspect.

I'm coming from a CS perspective.

If they are successful at a top university and leave to try other things (usually startups), then they will have no problems going back to academia. Sometimes it is even viewed positively.

For example, right now Carnegie Mellon has been seeing a flock of their faculty leave to go to tech companies or to do their own startups. Other universities would fight over a successful ex-CMU (or UW!) prof to come to their institution.

Depending on their level of success, they may not get back to a top 5 university, but some other R1 university would gladly take them!

I suspect CS is very different, but I don't know. When the dotcom boom started, many CS departments and universities seemed to start falling all over themselves, to get pieces of the action. (Whether for technology transfer or impact, or simply money.) I don't know whether Mathematics departments have had that kind of cultural upheaval.

I bet he could get tenure in a CS department if he tried.

(author of blog) If I were to attempt to return to academia, I would consider departments other than pure mathematics. E.g., the last place I interviewed at (about 10 years ago) was Emory University's department of Mathematics and Computer Science, and I remember being excited by computer science being part of the same department. I just checked, and see that it's now split into two departments: http://www.mathcs.emory.edu/site/home/

I would likely have a lot of trouble re-entering academia due to lack of recent academic publications; I've been on many hiring committees over the years in pure math, and we never considered hiring anybody who wasn't already in academia.

Could you say more about how noted professor marketability would work in this case, in your field?

My reasons for asking... I've heard mathematics academia has its own version of ageism. (In dotcoms, at least, it's easy to go from being a star who everyone wants to hire, to most people not caring.) Plus I wonder how it would be perceived: a history of seeming to want to do a startup rather than be a professor. And a university that wanted to be the new home of a prestige research&eduction open source software project would have to see whether the prior university's involvement would complicate things.

> Many, many schools would hire him with tenure if he decided to go back to academia.

Pretty sure this is false.

I'd say this is true. It wouldn't be UW nor at the same salary he currently has but, idk, Texas Tech, North Dakota State or some such place would think he would be a high profile catch at least in the next two or three years.

I worked at Wolfram Research and I have sideline knowledge of how hard it is to successfully commercialize a CAS. I personally would wait for NSF funding to improve (it is an unfortunate reality how funding is prioritized nowadays), and perhaps get resourceful about developing your system further in an academic setting- I’ve done similar work as a university student looking for a resume booster.

Brutal, but:


1. Do it -- You're ok with this failing and being locked out of academia -- then go for it! Maybe, if it fails, you'd be happy joining cloudera or another placing building notebooks!

2. Do it -- You're on track for paying living wages, both for yourself and future employees, say within 6mo organically, or with outside investment (from your post... you're still before then). Maybe live abroad for a year to further cut down!

3. Go back for a year -- Let it grow a bit more, and push any income you'd have taken to others who are earlier in their career

In both 'do it' scenarios, get ready to pivot/expand to areas that will do a better job of keeping food on the table for your employees families, and to keep failing at it until something works. If you can't stomache that, wait till you can hire someone who will, or keep it as a hobby.

I normally wouldn't offer to chat w/ someone over the internet, but I'm close-ish to this space, and a stage or so ahead wrt commercialization, so would be happy to do so

That you are asking this question.. then No. And you should bootstrap this with UW or carefully manage your obligations. There are many profs with a foot on both sides.

I would not leave until you've achieved escape velocity.

Looking at the full scope of things here it seems like it would be batshit insane for this guy to resign his professorship (on the economic calculus). Of course if he hates every day at the university it’s different.

Since he brought up he may have been in the wrong place being in the math department, did he ever actually try to switch to the CS department?

In UW's case, that would be even more difficult than usual because of its organization. The mathematics department is in the College of Arts and Science and the CS department is in the engineering school. These are practically like separate universities within UW.

Indeed -- CS is now the "Paul G. Allen School of Computer Science and Engineering", so it's not just a department. I never tried to move to CS, and I have no idea if there would have been a path to do so. When I first got hired at UW in 2006, some faculty in the CS department were very supportive of Sage, even giving me some temporary office space in one of their old buildings.

I would have tried to explore that option before leaving completely. I know two professors who have switched fields, one was physics professor who did work in string theory, switched to genomics and got position at a medical school. Another was a CS professor who actually went in to math.

You sometimes see CS professors who have degrees in pure math just browsing different department's websites.

Flip a coin, when it's in the air, you will know what side you hope it lands on.

He's already decided, and is resigning, as per the end of the post.

I know this isn't exactly related, but as someone interested in numerical analysis, what is the difference between working on Sage, Octave, and Julia?

Sage is primarily developed to support pure mathematics research (and undergrad teaching). Probably the main possibly unique thing Sage brings to numerical analysis is good support for arbitrary precision and interval arithmetic. Beyond that, Sage is written in Python, so it inherits the standard numerical functionality from that ecosystem, namely numpy, scipy, pandas, etc. Sage also has excellent support for Cython, since much of Sage is written in Cython (I even made up the name "Cython").

I was going to counsel that you retain your tenure but I guess you've already decided to resign. (If so, why ask?)

The rationale for my recommendation is largely captured in other comments so I'll be brief: Winter is coming (macroeconomically). And it sounds like cocalc lacks: a strong balance sheet, a scalable sales model, and a story that would interest outside investors. As such, I personally would view the company as lifestyle/earnings supplementation and not core sustenance. It is also synergistic with your day job.

TBH I feel like there's very little risk with quitting to work on a project that already has users.

It seems like he's a good enough programmer that he could always find a job at a big tech company, make a bunch of money and then quit once he has enough saved up. And that's assuming he doesn't find any paid users or get any source of funding which again is hard but not impossible or that unlikely given that he already has a product with users.

I think he'll be just fine and I wish him the best.

Job market for programmers will not always be what it is today.

Curious what's your reasoning behind this? If anything I see programmer jobs as one of the only professions that will survive the next couple of decades.

That's true. I bet we'll see many more different sorts of jobs open up as tech increases in complexity.

I took a number theory class from you almost 10 years ago, and it changed my life. So, thanks. :)

I am commenting because I have read parts of your algebraic number theory book and used Sage extensively. The book helped me but did not significantly change anything about my life. It is possible to learn the same topics by reading other books. I love many parts of Sage and it has changed what I am able to do.

It seems like the options are

tenure: a path where you make incremental progress on narrow problems with the benefit of low stress and an excellent standard of living.

company: a path where you get to continue changing the world in a meaningful way. The cost is high stress and a less certain standard of living in your current city.

You provide a lot of details about your situation but I don't think they are relevant. The details would only matter if the worst case outcome of either path was a life that you could not bear to live. Given your many accomplishments this is not a problem.

I think people should try to minimize the regret in their lives rather than maximizing their average daily happiness. It sounds like you would regret not continuing to work on your company.

I view economics as the gating factor. If, as you posted downthread, you don't have lifetime financial security, don't drop your tenured job.

This is like the preflight "put on your own oxygen mask before you assist others with theirs" instruction. IMHO, you'll be far less able to contribute if you're scrambling to pay basic expenses.

It seems to me that a more flexible university would be lucky to have you. "William Stein" was a household name among the number theorists I knew.

The subtext I get from reading your question is that you are overwhelmed with the support issues. Problems can arise at any time, and need to be fixed within hours. That's not a way to run any kind of product. If you can solve that problem (lower expectations for support), then you should be able to go back to your 50% position and still make progress on CoCalc.

I find it interesting that nobody has asked this question: what is your life situation?

Do you have a family? House/mortgage? Health insurance? You mentioned that CoCalc hires some engineers, some hat’s the burn rate? What do you need to keep working on it?

If you have got your baseline covered, William, i’d say go for it. Your heart is clearly on this path.

> Do you have a family?

Only a spouse, who has an excellent job.

> House/mortgage?

Yes, I bought a detached house 12 years ago in Seattle.

> Health insurance?

Yes, through my spouse.

> You mentioned that CoCalc hires some engineers, so that’s the burn rate?

Yes, we employ 4 people (including me). If we employed three instead, we would not be losing money...

You have a 10% chance of success even if you are accepted into YCombinator (honest feedback: I wouldn’t invest in your company). Stay a professor, and don’t regret not wasting time as an entrepreneur. It’s overrated.

Most important question: what will minimize your future regrets?

If you are a developmental neuroscientists then yes. I need your job. The job market is not great, as you'd know.

Should I reply to the title of a posting without reading it?

No, I probably shouldn't.

Applications are open for YC Summer 2023

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact