I agree.

I also think it is a matter of priorities. Other companies have dumped millions of dollars and thousands of man hours onto this problem. Highly unlikely you'd discover something new when you are small or alone.

I'd lay money down that gender and similar metrics would not produce anything meaningful and actionable. Of course, you could design this into the site, bit tjat would take a back seat to thousands of other design decisions.

any good neural networks lectures/websites? want to make a rubbish poker AI using NN but have no idea how to do it

Shoe, I meant to reply to this one sooner, but not on my phone. I also wanted to ponder it a bit to prevent a 1000 word essay:

The cannibalization on certain products are very low. In the case of the product with no description, a major consideration is how the person finds this product. In this case, I think it is likely someone used Google first. They probably found the brand and saw they could get the exact same product cheaper elsewhere. I would expect the user to have 5 tabs open and comparing. They couldn't possibly be buying for the information on the page because there is no other information. Assuming some sophistication, they would be able to figure out that this is the same thing and dive for the lower price, faster shipping, etc.

Another factor that is apparent, and reflected above, is specificity. Think of it this way, if someone is buying t-shirts, you'd probably see conversion levels like this: "Nike shirt" > "Cheap Shirt" > "t shirts".

To extend that to your dating site analogy, higher conversion, probably measured in emails sent, would be something like "Asains" > "Blondes" > "25 to 30 year old" > "Within 5 miles" > "My Matches"

Another factor is color. Obviously certain colors sell much better than other colors and end up converting better.

Unfortunately, I don't have traffic flow metrics to look at, but it would be nice to test some of my assumptions with it. In this case, traffic flow would be a final test to examine my assumptions. It is pretty key, but I've never used it as my first metric.

With that said, there are quite a few tricks we use that help out with sales and conversions, which I can't discuss here.

Those two in secession...fear the PC trolls imo

Numbers don't care about human emotions.

The color was referencing the product colors, not the people colors.

Is there a good resource to learn about memory management in javascript applications. I.e. this app I just built, I'm pulling in sometimes in the order of 3-500 objects off an API. If I paginate those locally, how much memory are those objects storing in the background? I don't even know if I'm in the ballpark with thinking that 500 objects stored in memory is a problem.

I also have a hard time testing stuff because I only own a retina MBP, and a super jacked gaming PC, which can obviously handle the stuff I build no problem. I'm using angularjs, and am mainly keeping track of the number of watchers on a page at a given time. I read a SO post a while ago saying that up to 3,000 shouldn't really cause a problem, but it depends a lot on how much stuff you're redrawing, and how cluttered it is visually.

I'm sure a quick google will reveal a chrome extension, and a number of blog posts on the subject. Just wondered if anyone could offer some insight.

ncha,

on chrome, shift + esc brings up a task manager that shows memory usage on a per tab basis. should give you a quick and dirty estimate.

Cool, thanks. Seems like it's between 100-125mb, which doesn't seem terrible compared to a lot of other sites I tested.

I suspect for most people bandwidth/performance would be a problem before the memory issue.

another option, which you've probably considered, is to pull maybe only the current page and the next page of objects, since your using pagination. every time the page turns, instantly make another API call in the bg to get the next page of results. this allows the app to feel instant to the user, but still avoids sending data they'll never see and using RAM unnecessarily.

Have you thought about running a virtual machine with a capped amount of resources? Virtualbox lets you easily set the RAM and # of cores for the OS you spin up.

You could probably get more in depth to cap the CPU speed inside of the OS. I know linux lets you introduce random arbitrary network lag too.

Example:
http://www.uponmyshoulder.com/blog/2...ions-on-linux/

I'm not designing the API, just consuming it. I'm not too worried on this particular app because it's an internal dashboard that's going to be used by ~5 people, all of which have modern computers.

But I agree, if I were writing the API myself I'd server side paginate it... maybe. I have a lot of table data with sorting/searching/etc, and that's kind of a pain to implement across server paginated data.

I'm not sure why people actually like Apple support. They cover things pretty well but actually talking to someone is a gigantic pain in the ass. I had to make an appt. to return a power chord. Then the battery needed replacing and that was another appointment just to drop it off.

****ing annoying.

Faster or slower than replacing Dell or Lenovo parts?

Touché.

Edit: I think my complaint is more that they have this highly advanced system for customer support but there's no differentiation between "I'm here to exchange this clearly frayed cord" or "I'm here to drop off my computer that you just looked at the other day" from "My friend gave me pictures on this frisbee thing but I want to look at them on my ipad."

I know it just looked funny back to back and I constantly struggle with PC folks :P

That's why I switched to a Mac.

I am reading a book that just came out in December, Professional Embedded ARM Development, that has an an interesting section on the history of the ARM architecture. I knew a lot of it but didn't know that Apple was one of the original members of the consortium. It is very good perspective on how things evolved.

How hard is it to implement Elasticsearch over a normal SQL database? I'm looking around and I'm afraid I'll end up tearing my hair out.

Depends on what framework you're using.

It was trivial to use elasticsearch and postgres together with rails without having to do much extra outside of what you would normally do with rails.

If you're not using a framework then you just need to make sure your ES index stays in sync with your sql DB. There's nothing special about it. You can write the glue code to this in real time or setup a cronjob to do it at specific intervals if you don't need soft real time updates.

You have a couple of approaches but the one I like is:

1. Use postgres as a primary data store and a single point of truth.

2. Use ES to index data that I want to search and facet on but I treat the data as not important. It stays in sync with postgres by reading from postgres. The ES index may also contain meta data not included in postgres for faster/better/richer query results.

I like this because then you are free to blow out your ES index on a whim if you want to tweak things that require building a new index. You can go with this approach even if you have 100s of thousands of documents stored. ES is really fast when it comes to building a new index and importing.

Edit:

Maybe I fail at life and you meant "over" as in a replacement? Then it's also really easy. You can use ES as a typical document store and query things.

So what is the use case where you decide you need elastic search rather than just creating a search API interface to your postgres database? Is it a lot faster? Are you just saving yourself the trouble of writing an API? What else?

Good answer. Thanks. Yes, I meant "on top of," not"instead of."

PostgreSQL does offer quite a few modules for searching which, when combined, is quite powerful. I'm sort of looking for facet search (which can be done with the cube module), fuzzy search (which can be done with levenshtein), etc., but was thinking ES may be more robust and easier to work with.

The plus is that I could use all native SQL, but the minus is that combining all those parts gets highly sporadic. Why shouldn't I use the expertise of someone else to solve this problem instead?

The more I use Foundation, the more I loathe it.

Still better than Bootstrap. :/

I kinda like foundation for naked projects. What's your objection?

I also like bootstrap 3 quite a bit relative to bootstrap 2. I've found a lot of good themes for it recently for various projects on wrapbootstrap, which helps me because I have no eye for design.

I like foundation better than bootstrap, but they serve different purposes at this point. I use the bootstrap JS a lot because they've all been rewritten into native angular directives.

I'm sick of all the silly defaults that don't make an iota of sense. I can get the idea that <p> is #666 (or something close to that in v4), but that is okay because you can modify all those strange settings on the download page.

And the download page screams "customize me!" since you can download all the things you want from the "kitchen sink" to plain ol' CSS, and this promise of customization is nice, but once you start using it, you either have to write 5 level paths to change minor things or you can't change it at all.

I'm left to ask a lot of questions:

-- Why is a "responsive" framework set to max wide-ass left and right white space, which is even sacrifices a ton of real estate on my 15" screen?

-- Why are the padding settings so damn large? Do I really want 50px separating each of my paragraphs? Why is the padding under <hX> tags so darn. Maybe I'd like stuff tucked underneath it. (this stuff is fixable, but still)

-- Why the **** are the padding setting so inconsistent. I want a nice responsive top-menu, but if I have one <span> or <p> element and one <a> element, they don't align.

-- Why can't I change the paddings on <span> and <a> without feeling like I just did CSS ninjitsu?

-- Why can't I have a search box that looks like this:



You end up with the search button with padding 0 and the search box itself with padding, 20, which means you can't get them aligned. Hell, this is even wrong on foundation's site itself! (this appears to be fixed in foundation 5)

-- Why is there a new version every few months and why should I invest the time to upgrade so I can have a feature that was possible in 2003?

-- Why are the default settings so damn ugly?

-- Why do I need a 7 gig file just to use 4 features?

So basically, you can customize it to some small point, but it after a bit more, there is nothing else that can be done. You get the easy stuff easy and the hard stuff still hard if you are you lucky, but oftentimes impossible.