We did a minor upgrade today between 11:00 am and 11:30 am Pacific time. For a change, we actually took the site offline during this upgrade. The reason for this was because it provided a good time to do a quick database recycle and optimization, and since we were upgrading a few core files (the code that runs the deep backend of the site), we didn't want to risk someone being on the site and doing a page load right as an old file was removed and a new one put in place. Although the odds are vanishingly small of something going wrong when that happens, it's possible - so why risk it?
This upgrade is what is called a "dot release," meaning that it's mostly very minor bugs, unreported bugs (things we've discovered that nobody else has), and two security fixes that, frankly, we're not going to tell you about. :-) (No, that doesn't mean we're hiding anything, that means that we patched two potential security holes that nobody had found yet. There's no reason to give hackers any ideas by talking about them, though.)
We're in the home stretch right now with the site. We have one major upgrade to an existing system, one major upgrade to a new system, and one surprise, all scheduled to be completed this summer. We've hired some new people in the past month, and they're working on other, new things that we hope will impress you.
Oh, and we still haven't had a single 502 error since we went online over a year ago. Just saying.
Or, Why Model Insider loves its international members!
With the rollout of the new casting calls, travel notices and directory system, we bring another aspect of Model Insider into the global realm and add geolocation to all of those systems. What does this mean? Simply put, Model Insider is the first truly global networking site in our industry.
We don't rely on ZIP codes or postal codes, making the site US-centric or Canada-centric. We don't have a huge database of cities that is not only missing many important areas but also suffers from inaccuracies and an inability to pinpoint your location if you live someplace not in the database. And we don't require you to jump through hoops to find your location or the location of others.
As described in a previous blog post, we've chosen to simply rely on the expertise and experience of Google, who have solved this problem with Google Maps. Every user has their location geocoded (that's geek-speak for "you tell us where you are") on a map. Every casting call, every travel notice, every studio listing ... all have their locations on a map. When you go to use any system at Model Insider, you can bring up a map and find what you're looking for.
Zoom in. Zoom out. Pan around. Filter your results to eliminate things you're not interested in. Expand your results to find those things you want. The capabilities are pretty-much limitless.
And while we've not been able to anticipate everything our membership wants to do, we have the ability to add new features and functionality as you request them. So use the new systems and get us feedback on what you think. We have developers just sitting around, bored, waiting to implement your ideas!
Okay, that "sitting around bored" part isn't true.
We had a major code rollout over the weekend, so I'd like to take you through it and let you all know what's new, what's changed, and our thought process behind it all. Before I dive in, though, let me note two things. First, these enhancements and changes are based on a lot of feedback. We get a lot of email with ideas and suggestions and we read and consider them all. With a nod that they are often contradictory (one person wants something and another person wants the opposite), trends do show up, and a good idea is a good idea. So we listen, and it plays a huge role in our planning. Second, we're not perfect. Many of the changes we roll out are evolutionary, as opposed to revolutionary. What that means is that if something doesn't work, we're not afraid to try something else. Perhaps that "something else" will also fail; we're not afraid to fail and try again if that's what it takes. That said, we feel that we've got a solid platform now, and we're working to make it even better.
So... what have we done? Simply put, we replaced the casting calls and travel notices with a much better system, added full geolocation services and expanded the system to create directories for events, studios, classified ads and services. We can create new directories (or remove underused ones) immediately, and we can also change the layout and fields of any of these systems on the fly. What that means is that if we missed something, we can correct that omission. Already, we're getting some feedback that studio listings are missing some obvious information, and we're going to add those in the next day or two.
The framework for casting calls and travel notices is pretty basic right now, but extremely powerful. The database can be searched on any field, the mapping system leverages Google Maps (meaning that we get anything that Google creates), and our system is designed to let us pull data from it in many different ways. Some of our ideas include promoting featured castings and travel notices on our search engine, TFP.me, and other web sites (both our own, as well as those we will be partnering with). You read that right: it's our intention to help you get your casting calls and travel notices seen. Part of a successful networking site is getting you the exposure you need to book those shoots. We have some serious ideas on how to do this, and we're putting those ideas to work. Now that we have a solid system, we start to turn our attention to using it. That's part of the benefit of designing Model Insider right - we don't have to spend a lot of time fixing ongoing problems and apologizing for the inconvenience.
So place a casting call! Place a travel notice! Advertise your studio or event! And then drop us an email and let us know what you think. There's no way we got this 100% right the first time, so please feel free to not only let us know what you like, but what you don't like. Is anything confusing? Is anything missing? Is there anything we put in there that you think should be removed? Remember, we listen!
And stay tuned... the next blog entry will be about the new top-level menu structure and why we did it that way.
We just rolled out the new directory system. This is a very short blog entry just to let everyone know that it's public and that we're running it through its paces. You could call it a public beta, but it's already been tested for functionality, so that wouldn't be quite right. What we're doing now is making it available and watching how people use it and asking our membership to give us feedback on what they like, what they don't like, and how we can refine the system and make it better. It's coded in such a way that rapid changes can be made at any time and with no downtime.
This new system runs our casting calls and travel notice system, complete with map-based geolocation, proximity searching and radius notification. It also supports our directory system which includes classified ads, a studio directory, services directory and events directory.
We can add information to any of those directories, as well as add new directories at any time, so we're serious when we say we want to hear from you. We've tried to anticipate as many needs as we could, but there's no way that we got this 100% right - but with usage and feedback, we can and will!
There will be a much more detailed blog entry later, outlining the features of the new directory system and answering your questions.
This article goes beyond the typical Nigerian scam, where a model or photographer (or anyone, for that matter) is told that they’ve been selected for a job, paid in advance with a fake check or money order, and asked to wire money from that fake check elsewhere. When the check is returned as fraudulent, the victim is out the money that they wired. In this more detailed article by Roger Talley, the author of “The Professional’s Guide to Modeling,” many of the other kinds of scams are explained. Often, someone might not realize that something is wrong because it doesn’t fit the pattern of the known Nigerian scam. Reading this article will educate members of our community and teach them that not only are there many different kinds of scams, but how to recognize them and avoid them.
Model Insider is pleased to provide this information for the community on behalf of and with the compliments of Roger and his web site, NewModels.com – we highly recommend that everyone, new and experienced alike, thoroughly read his site and encourage everyone to help new members of our community by providing them links to these sites and articles. With a little education, we can prevent a lot!
Well, it's not a new name in the sense that we're changing our name, because we're not. But I thought I'd make a quick blog entry to point out that you'll likely be noticing that many of our site copyrights and addresses now reflect that Model Insider is owned and operated by Insider House, LLC.
As with any business, we had to set up the appropriate legal framework to let us run the site, hire people, pay people, and do all those fun things that businesses get to do. We gave a lot of thought to coming up with a good name, and we think Insider House encompasses what we're trying to do. Insider, because that's the name of the site. And House, as in "publishing house." We see ourselves as a publisher, and as a company that enables you, our users, to publish as well. Networking is all about getting yourself out there, and what better way to do that than to publish your portfolio and network?
Hence, Insider House, LLC.
Of course, it's just a legal formality, and all this and $3.50 will get you a really good latte at the Blue Cow Espresso just down the street from our office :-) (really, they do make the best coffee in the Seattle area).
Odds are you didn't notice unless you were watching our Facebook or Twitter feed.
At 11:30 this morning (and that's Pacific Daylight time) I took the site into maintenance mode for an upgrade. To the end user, this has the effect of slowing the site down a little. I then disabled and removed two modules (our core dispatcher and commenting system, if you're curious). If you had happened to click on a blog entry or image that had comments enabled during this time, you would have noted that the comments were gone. This lasted for about 45 seconds while the new code was installed. Once that was done, everything would have appeared pretty-much the same.
We then validated the upgraded code and restarted the database. We didn't need to restart the database (it had been running without a reboot since March 8th), but I tend to be a little paranoid about these things. The site had downtime of precisely 24 seconds.
24 seconds. Yes, I timed it.
I then ran a suite of tests that verifies that the push was complete and correct, and, sure enough, there was one failure. There always is. One file didn't come over properly, and was causing some images to come up as 404 (HTTP code 404 - as in "unavailable."). I sent a quick Facebook message (which also Tweets it) to let people know and re-pushed the missing file. Caches were cleared, registries were rebuilt, and much to nobody's surprise, the 404s went away.
So what was in this update? A metric ton of stuff that you really don't see. Backend code that supports a major system upgrade that we're now getting ready to push out that will completely re-implement a number of features on the site based on user feedback (and, I should be honest, a lot of complaints). We listened to those complaints and redesigned a number of things that our members said just weren't working the way they should. I'd clue you in to what they are, but we want it to be a surprise. Once we push this out, there's just one more major system that we're going to re-implement, and we're done with the second release of Model Insider. Call it Version 2 if you will. At that point, we feel that we have the basic features and functionality that all modeling and photography networking sites should have.
Then we REALLY get to work. This spring and summer, we have two overriding goals. The first is to introduce a few new features that we're designing that will blow you away. No other site has them. No other site has even talked about them. We think that they're going to take you completely by surprise.
Our second goal is a redesign of the look and feel of the site. Our design could be better. Hell, I'll be honest, I designed most of it, initially, based on a commercial template that I like. I had Daniel, our front-end developer, do some clean-up of my original cruddy design and he's made it more than workable. But the truth is that we need this place redesigned to be easier on the eyes, easier to use, and simpler in terms of the layout and flow. Now that we have the core system functional and useful, it's time to put some energy (and not a small amount of money) into that.
We're here for the long run, and we're investing in the site.
Oh, and did I mention that we had a major upgrade today? Would you have noticed if I hadn't said something? Because, you know, I can fake some 502 errors if it will make anyone feel more comfortable that we're actually doing stuff here instead of just sitting around, looking at your pictures.
You let me know.
As many of you may be aware by now, Model Mayhem has had a small “glitch” in their system. In this case, they’ve started compressing and stripping all uploads, irrespective of size. Please see Pat Yuen's blog at http://blog.patyuen.com if you’re not up on what’s happening. Go read it now. I’ll wait.
Back? Good. By the way, his blog is an excellent source of ongoing information, and we highly recommend that you bookmark it or add it to your feed reader.
So now that you're aware of the issue, here’s how Model Insider works, just so you’re all informed and aware.
Model Insider allows free accounts to have images up to 800 pixels in length on the longest side. There is no differential between horizontal and vertical. Paid accounts can have larger pixel lengths, 1000, 1200 and up to 1600 depending on account level (with a nod that other than the short permanent sale we had last month, we've not started offering paid accounts yet. We likely will just as soon as we're comfortable that we can offer the service that a paid account should get with little chance of breaking. We'd rather have things working right across the board before we accept your money. But that's us).
I digress. If an image is uploaded at or below this limit, the image is simply copied with no adjustments made whatsoever. If an image is longer, it is re-sized. Until today, we used GD2 to make thumbnails and re-size images. One downside to GD2 is that it compresses at a slightly lower quality than other libraries and it also strips out all EXIF data. Until now, we’d dealt with this as a known issue and nobody has complained (much) because the clear workaround is to simply re-size your own images before uploading.
Based on the current issue elsewhere, however, we felt that it was wise for Model Insider to learn from the current mistake and although we did not, do not, and never have had this problem, we have ensured that we also will not, going forward. This afternoon we did the work to change our graphical library from the less-than-optimal GD2 to ImageMagick, which is better, faster, and does not strip EXIF information. It also generates higher quality images, at the expense of size. That means that Model Insider will have to store more bits and will be paying a little more for bandwidth.
This is something we're more than prepared to do, without a second thought. As a site for photographers and models, if we can’t offer the highest quality possible, what’s the point?
This was on our roadmap to get done at some point in the future, but the recent turn of events has made it apparent that doing it now might be wise. Anticipating that we might want to improve things going forward, we have the ability to “reset” ourselves. We store all original uploads, even if they’re thousands of pixels across. We store them for this very reason, in fact. Right now, a full backup is being run on all images in our database. Once that backup is complete and verified, we will be regenerating all display images from originals using the new library. That means that any display images that previously had the EXIF stripped will have it restored. This regeneration will take about a day, but you will notice no difference. Images will be replaced, in-place, in real time. Of course, if you uploaded your image without EXIF data, we won't be adding any for you. Perhaps that might be a funny April Fool joke, but it's a little late now. :-)
All new uploads as of right now are using the new library, so there will be no gap.
And nobody has to wait until “normal business hours” in California for this to happen. We did it today. As we've said, Model Insider isn't run by some corporation just looking to make a buck. We run this site because we're photographers and models ourselves, and we are part of the same community you are.
Today is my birthday, but Model Insider gets the present (and before anyone else calls me an old fart, I'll have you know that my age corresponds to the element with the lowest atomic number that has no stable isotopes - so not only am I a geek, I'm also not that old!)
Many web sites suffer from the problem of reliable email delivery. From that initial email at signup, asking you to confirm your very existence, to the notification that you have a new private message, or that someone has commented on one of your pictures, or even that there is a new post in the forum (though some sites don't have that), it's never any fun to have those emails go missing.
There are any of a number of reasons why this happens. Perhaps the notification system on a site is broken and simply not generating the mail. Perhaps there's so much that it's backlogged, as the site's email servers can't keep up with the volume. Perhaps the mail was rejected as spam or, worse, simply ignored and trashed without ever being presented to the user. What is a site to do?
The answer is simple: fix it. At the end of the day, reasons and excuses don't matter. Results do.
That's why Model Insider rolled-out its new internal email system today. This is a system that you, the user, will not even know is there. And since you've been getting your notifications reliably since day one, you shouldn't even notice that we've made the system even more reliable and prepared for the kind of scaling necessary as we grow.
First, guaranteed generation: our new system not only generates notification email, it also keeps a record, in the database, of every event that needs a notification. When the email is generated, that record is updated. No email? No update. And then the next time the system goes to see if everything is running properly, it sees that a notification wasn't generated and it tries again. Three failures and a human is notified that there's a problem.
Second, volume scaling: we've implemented an advanced queue. That's geek-speak for the fact that we can generate notifications as fast as necessary and then send them off for delivery as fast as our system can handle. And that's pretty fast, considering how we've solved the third problem.
That third problem is delivery, and that's the big one. How do you ensure that the notification emails actually get delivered? At worst, how do you ensure that you've done your best to get those emails delivered? Well, we could hire someone to call every ISP in the world and work to get us on their "whitelists." We could constantly ask our users to sign up for gmail or hotmail accounts because AOL likes to filter notifications as spam, and then apologize for the inconvenience. Or we could simply tell the users to check their spam boxes and leave it up to them to wonder if we were sending the notifications in the first place.
We chose to take a different path: we hired an expert and partnered with a company that makes it their sole line of business to ensure that email gets delivered properly. As of today, all of our outgoing validations and notifications are being sent through commercial-grade email servers by a company that verifies that their customers (that's us) aren't spammers, that the email is legitimate and genuine, and then they work with ISPs and other services to make sure that their email gets delivered. They track the email from submission to delivery so they can spot any problems along the way and resolve them. And they allow us to track them as well so we can report to you if there are any problems. If an ISP starts rejecting our email for any reason, we will likely know before you will and address the issue immediately. If your confirmation email bounces, we'll know, and if you submit a helpdesk ticket, we can check before getting back to you, allowing us to see the problem instantly.
From a business perspective, what we did was realize that it made very little sense for us to reinvent the wheel and try to become email delivery experts when there are companies out there who make it their business to provide this service. They participate in whitelist programs with major service providers. Their agreements cover AOL, Juno, Netzero, AT&T and Comcast, just to name a few. They constantly monitor email blacklists to ensure they're not on them (and when they do get on, as will happen, they deal with the issue immediately). They monitor their customers for spammers to ensure that their networks are never used for spam (indeed, we had to answer a battery of questions to qualify for their service).
So what's all this geekery really mean?
Simple: we don't like it when things don't work. We insist upon providing a web site and a community where things work. And when they do break, we don't rest until we've figured out the problem and fixed it. We did that with search, then we did it with the global location problem, and now we've done it for email delivery. Wait until you see what we solve next.
Please tell your friends about Model Insider. We're here for the duration and we genuinely appreciate your support.
Being able to be found is important. Finding other people is just as important. These days, there isn't a networking web site out there that doesn't have some kind of robust search system, to some degree. The one thing that all of these systems have in common is that members need to tell the site where they are. You can't tell someone how far away they are from someone else if you don't know where both of them are. But how do you let users specify their location?
One solution, buying a huge database with every city in it, is somewhat nonsensical. First, those databases are expensive, if you get one that’s actually decent. If you go for a cheap database, you end up with gaps in your cities, at best, and incorrect information, at worst. Can you imagine deploying a database of cities around the world and including a city that was sacked and burned to the ground over a thousand years ago?
There are good databases, too - but at over $3,000 for a subscription to a worldwide database, that’s an option that just won’t fly. That, plus those databases still don't have every city, some are named wrong, and some that are large actually pad your radius searches by up to 30 miles! Combine that with constant maintenance, and it's a nightmare.
Like most sites, we chose to do a latitude/longitude solution, but without the database. Until now, we mapped zip and postal codes, but for the rest of the world, had to presume that our users knew (or could figure out) their latitude and longitude. Yes, it's not rocket science, but we didn't want to have to make that presumption. We wanted to make it as easy as possible.
So we chose a mapping solution and leveraged the power of Google, once again (as we did with our TFP.me search engine). With this solution, you simply put in as much of your address as you want, anywhere in the world, and click “Map It” to bring up the Google map. Place the marker on your location, close the map, and hit save.
Anywhere in the world, named or not. With incredible accuracy, up to 12 decimal places.
Once again, we leverage the power of an existing solution, inexpensively, and better. That's the whole idea behind Model Insider - we're learning from years of experience, figuring out what works (and what doesn't), and implementing it. One feature at a time, we are consistently becoming the most feature-rich and most reliable modeling and photography networking site on the web. And with what we have planned over the next few months, we think we'll really knock your socks (or other items of clothing) off!