Np, up until a few days ago full text search was just some black box to me. I knew it was useful but I never worked with it. Searching is just scratching the surface too. You can do all sorts of crazy real time analysis stuff too.

I had an urge to implement a decent search on some app so I dove in. It's actually really cool. Once you actually see your data returned with highly relative results you just want to fist pump because it was so easy to setup.

Only downside I see is if you have a barebones VPS it might be tough getting everything on 1 box. ES and solr both run on the JVM.

i don't work with the search guys a ton so sorry for my misleading and/or incomplete answers. looks like shoe has filled in for me nicely .

i can't share specifics for a variety of reasons, but yes, it is a big cluster -- QPS in the hundreds.

I'm sorry, did you just suggest that there are any cool, or even uncool, open source or any programming related projects that amply describe how you use them?

My observation is that you ask a question on something then they sort of call you an idiot for not getting it.

After much deliberation on what to do about the Salesforce situation, I decided to strike it along with all the other CRM solutions and took a look at ticket trackers instead. Turns out that a CRM really doesn't equate the company's business model.

I evaluated a few of them and found http://www.otrs.com/en/ to be the best of the FOSS bunch. Or rather, the one that is least likely to cause me a bunch of headache to set up. It is actually easier to use than Zen Desk, and certainly less cluttered and more intuitive / powerful than Salesforce in the raw: there is real working email assignments, email locking, and of course, a way to make the UI look better.

Of course, to really make it function, one must use Perl. I'm going to suggest they call a professional for that stuff.

well actually i misspoke, it's commercial companies where the phenomenon happens most, which is why it's so surprising.

programmer superiority complex is another beast altogether.

i've always found the ruby mantra MINISWAN (matz is nice and so we are nice) to be revealing on that point and sad and obnoxious -- not because the sentiment is wrong (it's great), but because it has to be stated as this huge exception to the rule, and the underlying assumption is, "well, we're all *actually* a bunch of smug dickheads who would love to be rude and clannish, but since matz isn't like that, i *guess* we can not be like that too. but man, it's going to be hard!"

Hi Guys,

Can anyone recommend a really simple CMS/User Management system?

I am looking for something very basic which supports user registration (with some level of customization), and a user 'account' page where a user can change their password etc.

Drupal, wordpress, joomla etc are all way way too convoluted for what I want.

Dont mind what language/technology is used, but it needs to be able to run on a linux box (so no .net), and would be ideal if it supported MongoDB (but not essential).

Thanks!

i hear dave really likes sales force dot biz dot see oh dot you kay.

also, Mono you insensitive clod.

Not sure about this. I'm running two sites on one box both based on JVM. The JVM has excellent JIT and garbage collection, so I think that while the load up is a bit slow, the footprint once its running is relatively small.

FWIW, if you haven't had a chance to implement document mappers and reducers, you ought to give it a try. Its a surprisingly simple and elegant algorithm. Here's a working javascript and python implementation: http://jsmapreduce.com/

There was a discussion on HN recently about how having a box that said "sign up" with no real description of what you were signing up for was an anti-pattern. The follow-up to this point was interesting because several posters were quick to point out that their data showed that this was the best way to improve sign ups, conversions, retention, and thus, profit.

The same goes with the propagandist ambigu-speak that pervades most of these sites. The modern marketer isn't just tossing out ideas and hoping for money to follow. No, he is analyzing and crunching, and counter-intuitively, this sort of meaningless writing gets the highest conversions.

Yeah, you and I are more prone to dealing with "plain English," but the ugly fact is that the style here creates more income.

Data / intuition stuff quite often never really works out like one would think (poker analogy?). I was creating ads for the company I worked for. Now, the intuition is that image ads create more clicks (this was PPV, so you had to get the CTR up), improve "branding", and generally worked out for all sorts of reasons. These were more expensive than the all-text ads, so there was more confirmation. I ran image, image / text, and plain text ads. I discovered two strange things: First, the pure text ads out-performed the image ads by 10x, and second, long copy out-performed shot copy. The intuition is that people prefer images and don't have time to read, and thus slick graphics and less text is better, but the data suggested the complete opposite!

As an aside, that experience molded my opinion on the present trend to stupefy everything into symbols. Yeah, the flat-designers think that is the best, but I think that if they really ran the data, they'd find that text is far superior. The default view of Gmail is an excellent example of why removing text is a terrible idea.

So, there was this competition called "Clojure Cup," where teams of 1 person to 4 people competed to create the best app they could in 48 hours. The submissions are in and ready for public vote: http://clojurecup.com/apps.html

Codenotes and a few others look pretty cool. Overall, they seem a bit buggy, not starting, or not explaining what they do, but for 48 hours, it isn't too bad.

good post, dave. i agree with all of it.

i'd add that i'd be much more okay with the hyperbole and ambiguousness if there was also, maybe even on a linked to page, an actual explanation of what the product is. but maybe people that "let's send you text messages to friends" detracts from their self-image of "a social web in the palm of your hand that connects you to the most important people in your life. instantly."

oh btw i think we could have an entire thread about the conventions of startup web copy post 2011. it would be dead-simple!

I think that is why I'm starting to get attracted to those long-scroll screens that are becoming popular. It is quite effective when you think about it intuitively (and there is plenty of data backing this up: source), for the same reason you state. You figure the top-level is going to grab those who are most interested anyways and can be easily convinced, then those that want more information are going to scroll for more information, then, say you are buying a Nexus7, and you want more information, there is no clutter that detracts average user from the message, but there is easy access to specs if you are interested.

I think that, if you are really trying to get 'em all, then I think that the tall pages are the way to go, but I think that there is that opportunity loss that companies are considering as well: why should they pay bandwidth for free customer samples, etc?

I think that part of this trend of ambiguous copy can be attributed to people like Simon Sinek. This video circa 2010 and many others of his are among the most watched videos TED ever released, and I know from my earlier dealings with companies that the influence of his thoughts is massive. A whole cottage / business consulting industry has been built around his ideas. One of the key elements of his ideas is that we, as consumers, don't buy stuff but ideas. I think that the key point that is missing from many of these copies is that they still are inward focus, not outward focus. Apple, whom Sinek uses as his example here, never really talked about how great they are as a company, but how the user is empowered by their products. I think Apple used a mix of new-school Sinek thought and old-school Carnegie thought.

Can't remember if I or anyone else has posted this here, but I thought this was a good talk about the bolded insight:

http://businessofsoftware.org/2013/0...t-development/

Her idea is that what we really want to buy is a new a vision of ourselves, a different and better life. When you go all the way to the bottom, that's what it's always about. It rings true to me.

Just watched the Sinek video you linked. I see what you mean about his influence on startup speak and web page copy.

It's funny, he's a good speaker, and I agree with him: I *do* want to use tools and software developed by people who share my aesthetics and philosophy of design. That is the true root of the decision.

But people go wrong when they draw too literal a conclusion from this and focus on describing the "why" of their technical product. At least in my case, your ability to speak plainly and clearly and specifically *is* the proof that we are like minded and share the same "why." Whereas if you start by explicitly telling me your "why," then I know it's not the same as mine.

I'm just defaulting to Kant when it comes to that stuff. I'm fairly up to date on behavioural economics and know the irrational behaviour patterns and how they can be "exploited" but at the end of the day even if I know they'd improve the bottom line I'd refrain from using stuff that seems abusive/where I'd feel tricked if someone explained it to me.

I consider this part of business ethics.

Also...
Brag: 1.5/2 conference papers I submitted have been accepted (one was rejected but accepted as a shortened extended abstract so I'm counting that as 0.5)
Beat: No major conferences, Finland/Slovakia will be cold during November, not all that happy with the papers actually kinda rushed etc.
Variance: Might link PDFs after the conference, 0.5 one on is on our didactic concept for AI, accepted one on the architecture of our game server (Prolog+Java/Tomcat+HTML/CSS/JS+PostgreSQL)

Good thoughts here.

If you are not already familiar with the way Google (and Mayer specifically) views design, this is a good read: http://www.fastcompany.com/1403230/g...designers-data

Perhaps the best one?

relevant link and keeping the troll alive!

So I think I've found a new low in software defaults. I "upgraded" my phone to iOS 7 last week. After missing some important calls, I dug around and found out that the "Do Not Disturb" setting was changed to be on by default. The "Do Not Disturb" setting suppresses most incoming calls and messages. So my phone defaulted to not taking calls.

Re: dark patterns - Is it possible to quantify short-term conversions vs. long-term, particularly for things like ad-based sites where long-term users are a big deal? I have two Google+ accounts that I didn't even want, yet I go out of my way to avoid Google's ads out of spite because of this situation.

I suppose being vague in your signup process is a different matter because it infuriates fewer people, but I wonder if anyone has taken numbers on "I'm looking to jump ship the second someone who isn't a jerk offers a similar product" vs. "this is a cool thing made by cool people" or if it's even possible to do so.

Re: flat deisgn - Is there any data or even a single usability expert's opinion in favor of it? Not a rhretorical question, it just seems to me like a sin committed by graphic designers optimizing for screenshots and not users, but I'd be interested to see an opposing view.

I have a rails data modeling question:

Let's say you have a user and this user was created with devise.

Let's say you want the concept of an admin and a client to be applied to the user, this seems straight forward enough with roles. A user could be an admin or client just by setting their "role" column.

But what if you wanted to have a relationship between the 2 such that:

An admin can have many clients
A client can have 1 admin

I was thinking about doing some type of self ref solution where I introduce a third model which has through "project" because the thing that links the 2 is a project.

The problem I see is a client and an admin's fields will be much different. An admin might only be the login details + name while a client will have a ton of stuff like an address, company name, billing information, etc.. I'm not even sure yet but it will certainly be more than an admin.

How would you go about setting this up in such a way where it's not ultra complicated (STI) while still working off the same "user" model that IMO should only contain the necessary stuff to login and probably the role? That is question #2 kind of.

Thoughts?

summary:

benefits not features

Thought about #2 a bit more and maybe I could keep the user model super slim (id, email, password, role, etc.) and then each thing I want to add like an address or billing info would be their own models with a simple "address" belongs_to user and "user" has_many addresses?

I'm afraid this might lead to a billion joins but I guess it doesn't matter with fragment caching?

I'd do this. If the columns aren't going to overlap a huge amount, I think you're better off from a maintainability standpoint creating different profiles. It'll create some joins, but that's not the end of the world imo.

It also might make sense to just have a little duplication and create a client model and an admin model.

Just depends on if you want to avoid a ton of joins through the respective profiles, and just have two different user models. A lot of people recommend having a separate admin model, though I usually just stick to a roles_mask or admin boolean.

Thanks for the feedback.

The columns in this case will not overlap much all, there might be 1 or 2 common fields. My plan was to fragment cache as much as possible and I'd also be using elastic search to index everything that'll be searchable/facet'able.

Clients won't be able to register either as they'll be created in an admin dashboard. Admins won't be registered either, I'll just setup a default admin account in a seed and they can always make more admins if necessary.

Clients and admins will both need the same type of account related maintenance/features though like being able to reset their password, track sign ins, lock out protection and various other common traits. They will be doing that through the same UI so they could share controllers/views.

What benefits would separate models give me btw? In the future there's a small chance I may introduce a third role whose privs would be only being able to fully manage 1 or 2 resources but other than that they are basically mini admins. Does that change your decision?