This would be great for CERN. We do a lot of software development, but most of it is very bespoke, and you are often not going to get an answer on stackoverflow. Also there are confidentially requirements (it would be great if you could tag a post e.g. ATLAS-internal, CMS-internal, and then only people from your experiment could see that post). We probably have tens of thousands of members, so I think we have the critical size needed for such a project. Finally, I can see it not only used for programming, but also for physics, statistics, analysis questions in general.
Currently, we are mostly using old school mailing lists, and it can be hard to find some information you are looking for.
I would sign up, and prod our computing department to check this, but I'm quitting next week :-(.
I don't understand... why would you want to keep your discussions private to each experiment? Certainly people from other experiments or from other research groups could have very similar interests.
Moreover, CERN is funded by the taxes of us european citizens. Does this research really need to be kept secret?
If you study all your life for something, you want it kept held close to you as possible until "the right time" to extract maximum benefit/glory/$$$/fame etc. Wouldn't you?
ATLAS collaborator here, I just signed up :) We already have several un-official slack teams, some of which have gained a bit of traction with ~100 users. It seems this might catch on, even if CERN/ATLAS mgmt doesn't officially embrace it (which I suppose they would do via SO Enterprise, anyway).
I wonder if there had been some efforts to index all of the mailing lists into something like elastic search? Would elastic search help when a search's phrasing doesn't quite match phrases that have been indexed? Are there other solutions?
We are using a search solution, which I think is provided by Microsoft (http://search.cern.ch , I find it looks like one of these fake search engines when people on TV browse the internet ;-)). It works pretty well to index the mails, but could always be improved of course.
The bigger problem is that nobody wants to sign up for the traffic heavy support mailing lists. We actually have shifters who do that and try to answer the mails. If there was something that is "pull", not "push", or if I could just watch certain tags, it would be easier to participate.
Right, we are more worried about competing experiments seeing our data and discussions. (There may be other concerns, but I'm not a legal expert so I can't comment on that.)
The most compelling reason for confidentiality I've heard is that if another experiment sees our work in progress, it would influence them, and the results would no longer be independent. If you go back and look at old measurements, you find that they were often biased towards rumored values or other experiment's measurements.
(There are also less then convincing reasons, many people fear releasing "unfinished" information to the public would make a bad impression and risk funding. I'm personally inclined to let the public see more how the sausage (err, science) is made. People get the wrong impression that we turn on a machine, get the result, and write it down for eternity into a textbook, but in reality it is always a long an iterative process with many mistakes along the way.)
That, and one gets usually chastised when not finding a duplicate via this search box.
I always check Google results, then look at the yellow suggestions box for earlier posts and sometimes end up with a duplicate (rarely but still).
This usually ends up with a "closed as duplicate" (which is fine) but also sometimes end up with cynical, passive aggressive remarks on Meta.
I love SO (and SE in general). It helped me to understand and discover things I was not aware of.
But Meta is an abomination, full of people with ego issues and heavy complexes. I abhor this community, which is a shame as I stopped to file bugs in SO just because of the toxic environnement there.
PM on Stack Overflow Channels here. You make a valid point. :) And I can assure you, the success of Channels is predicated on a solid update to the way we do search (as well as better integration with tools you might already be using, like Slack).
> I don't know anyone who actually finds the answer via SO search.
I had to load up the site just to look where the search box even was. Not that it is hidden by any means, just that my eyes follow google -> topic title -> green arrow -> scan other solutions/comments and I had no mental reference of their search bar in my head
SO search never seem to find the answer I'm looking for, but once I start writing a question it almost always ends up finding a similar (answered) question. So their search works but perhaps it needs lots of input
Really not a fan of the "free for now, we'll give you an idea of pricing in the future". I understand that the team at StackOverflow probably doesn't have an idea on pricing, but it makes it extremely difficult to jump on and try the product.
I guess in that situation, our team should just wait until it's out and public. But a lot of these "free betas" go on for months or even years. I know this isn't always the case, but it almost feels like at times, the goal is to get users dependent on the product before telling them how much you're going to take/charge.
In my opinion, this is a tactic used when a company has less than maximum confidence in the product and want to test the water. Ie. It could be a great success and they will charge you $$$, or a complete failure and they will drop the product, or a "just ok" success, maybe due to the product itself being "ok". Even then best of these outcomes is not good (considering the risk of investing in the product early), so if they are not confident to charge 100%, I would not invest any significant time/resource into something like this. I'm out.
I agree, it's better to make a "beta price" or "pioneer price" that you honor, when I see "Free for now" I worry about investing effort in a product that the company does not have enough confidence in to put a price on. Especially one where I may be storing what is private intellectual property and trade secrets. I would much prefer to see a price and a commitment.
I see what you're saying. The honest answer is that we just really need to focus on making sure the product works well before anything else.
I really can't speak with confidence about pricing, because we still need to do discovery on what will work best. But what I do know is that it'll likely continue to be free for many people with some premium features that people can pay for if they want.
And if not, the data you create is yours and you can export it and bail!
Thank you. But when you say 'export it and bail', what do you mean? Is it exportable in a way I can fire it into something else? Is there a definition of 'exportable'?
This has already existed in the form of Stack Overflow Enterprise [1]. One major difference is that the company I was at managed and maintained their instance of Stack Overflow (it involved a lot of work). I wonder what else is different (or maybe this is just a replacement for Enterprise?)
For context, I'm the GM of the Enterprise Team at Stack Overflow and have been working on it since we launched our full Enterprise product about 18 months ago.
Enterprise is intended for large teams (at least 500 tech staff) who want to have their own completely isolated and standalone Stack Overflow community that can be run on-prem or in a private cloud. Enterprise deployments have full control over the system and also get support from our Customer Success team to build up their community using all the lessons we've learned in 8+ years of building communities. As part of this we have all those features you'd expect like integration with SSO, audibility, massive 50 page contracts, etc
Channels on the other hand is meant for smaller teams (all the way down to just 2 people) who want to store their own knowledge (privately and securely) alongside the the public knowledgebase at stackoverflow.com. Any size team will be able to just walk up, put down their credit card and instantly be sharing knowledge with each other.
So the two products are very complimentary to each other, just depends on your team size and exactly what you're looking for feature wise.
> store their own knowledge (privately and securely) alongside the the public knowledgebase at stackoverflow.com
I know this won't go over well here, since many depend on Stackoverflow for their "programming skills", but this sounds about as believable as Google pushing companies to store their secrets "privately" using Google Docs.
I'm not convinced by the free beta. I cannot ask my team members to start using it and add knowledge into this system with the risk that we'll have to migrate everything back once the costs are announced (and we'd loose time (= money) in the process).
I had a side project which was basically this (at least this was my [company] for [x] summary of the project) a couple of years ago. I always wondered why StackOverflow weren't doing it, and after years of wondering this I just built it myself.
I tried to publicise it a bit (not a lot, in fairness) and didn't generate much buzz. I even speculatively applied to YC with it, with the beta written and running, and got no call for an interview even. After that I lost enthusiasm for the idea, and moved on to other things. Seeing this has reminded me of it.
What do people think of this? Useful product? Would you pay for it?
It's a useful product. I'm actually surprised that the state of Stack Overflow clones isn't better. Although much of the value of SO comes from its knowledge base, the UI and organization and administration is nice to have pre-built. When I was looking around the top choices among open source/free were askbot (https://github.com/ASKBOT/askbot-devel) and osqa (https://github.com/dzone/osqa). There was also a plug-in for Confluence (Questions) and Answerhub (seemed to be the pro/paid version of OSQA). But none of them were as nice as stack overflow, which was surprising to me.
I think it's very cool, especially for large teams at big companies, but I'm not sure I can see my company using it. They seem to prefer home-grown solutions for lessons learned and other knowledge bases, because there is inevitably some special feature that someone considers necessary.
Then, of course, you end up with huge exported spreadsheets in every single workflow. sigh
Q&A tools (or features for existing community/collaboration tools) are an common request in the enterprise space. Virtually every large organization has some kind of knowledge manager either trying to build, or manage tools for this sort of thing. It's hard to understand why you didn't get a response without more context.
Had a team running on Askbot [1] for a few years, but nowadays we've been running on Questions for Confluence. I feel like most Atlassian shops will probably default to that route.
Stack overflow may offer lots of fancy features, but, in my opinion, the essence of what they do should be easily achievable by any competent group of engineers. Genuine question - I don't get it, what does this bring to the table? A voting system?
Microsoft built several of these Q&A systems that are "easily achievable by any competent group of engineers". They never really took off until they got Stack Overflow Enterprise. So, I guess it isn't really that easy.
Note: I'm a former Microsoft PM and currently a Stack Overflow PM. My POV comes with bias.
Can this do what SharePoint and private wikis can't? My team has tried both with varying degrees of success. In my experience it's always just "easier" for people to pester those who should know the answers.
Come over and ask me something I know is in the wikki? Let's sit down and search the wikki together. No match? Then let's type up a result together and enter it for future use.
You can't fix a bad process with a magic app, but a magic app can help a process work much better.
Both of those are much more "free form". You have to put effort in to get some value out. That's before you think of culture, and for it to be the norm for people to use those tools. SO has structure for Q&A which solves the technical issue, but it still doesn't solve the cultural aspect.
Does pestering being "easier" reflect a deeper problem? Maybe, maybe not, but in a lot of places people get annoyed having to answer the same question over and over. Committing answers to writing also helps clarify them and avoids someone walking off with institutional knowledge. In theory, that improves things.
>In my experience it's always just "easier" for people to pester those who should know the answers.
Your intuition is probably correct. It's interesting why private corporate wikis and other "knowledge management"/"corporate intelligence" systems fail to live up to the vendors' promises while public crowdsourced volunteer efforts like wikipedia.org and stackoverflow succeed. They have different social dynamics that software tools don't really solve:
1) Venn diagram intersection of typers & searchers is too small a quantity. The # of employees who will type out quality answers (typers) and # of employees who will search databases for answers (and attempting different keywords/synonyms when they initially get "0 results found") is too small to make internal knowledge databases flourish. Yes, the % of contributors[1] to wikipedia and stackoverflow is also less than 1% compared to lurkers but since that 1% in absolute numbers is much higher (e.g. 100,000+), that's enough to keep up the quality content on those websites.
2) out-of-band questions/answers that happen outside of the Q&A database. E.g. Bob catches Alice in the hallway as she heads out to lunch and asks her why WidgetX has foobarY and she can answers it on the spot. Neither are going to disrupt their motions so both can go sit down at a keyboard and type out the same Q&A again. Same situation with questions popping up on Slack or conference calls. Compare that in-office dynamic with remote Stackoverflow contributers[2]. You don't have those experts walking around your hallways so you must use stackoverflow to type out the questions. There are no out-of-band communication channels. This constraint adds to the quantity of content.
3) secretarial/administrative/housekeeping curation. Much of the usefulness of wikipedia & stackoverflow is the ability to ask questions in different ways with different wording but still end up at the optimal canonical answer. This happens because the community continually ads relevant "tags" to questions, flags duplicates (that had different wording) & point them to previous answers, and adds comments to old answers to signify they are obsolete. This makes the cross-referencing index of those public websites very good. In contrasts, corporate wikis usually have no full-time admin to continuously enhance the cross-references. Therefore, if a new employee doesn't know the exact magic words to ask a question, they'll get "0 results found".
If the underlying problem is people dynamics and incentives, it's very hard for software tools of any kind to solve that.
At one of the Big 4 firms, managers tried to incentivize contributions to the "knowledge database" by tying it to annual performance reviews. Guess what happened? At the end of the year, the database gets flooded with crap articles (e.g. Powerpoints and proposals for specific customers) uploaded to the db to get that requirement checked off. That type of useless content would be quickly deleted by admins on wikipedia and stackoverflow.
The combination of all the factors above results in corporate wikis being either 1) a barren wasteland or 2) a dumping ground for useless content.
What were some of your deciding factors? Does that forum software have modules or configurations that make it effective as a Q&A focused site? How does Flarum compare to say Discourse?
Currently, we are mostly using old school mailing lists, and it can be hard to find some information you are looking for.
I would sign up, and prod our computing department to check this, but I'm quitting next week :-(.