BrandonPollet.com

Hey everyone. As you may notice I haven't updated this blog in quite a while. Most of my writing as of late has been over at f5games.com. If you're interested in what I'm working on please head over there.

A special thanks to my friend David for putting this together for me. I need this on a t-shirt.

Here we are, only about 8 days away from my original deadline, and nowhere near a finished application. There are multiple reasons for this; my deadline was unrealistic, I underestimated my available time, my family and I are moving to a new house, and the list goes on. But, while the original goal will not be achieved in the desired timeframe I still have made great progress over the last few weeks.

I was able to get a fairly decent 'design document' written up, which is very different from the design documents you are probably familiar with but it works great for me. I have also been able to create html versions of the main application areas. I've come up with a rough overall design, mapped out most of my intended functionality, and even created a system to test out market demand for my product. All in all, it's a pretty good amount of work for a few weeks time.

The picture that accompanies this post is actually from one of my html mockups, and it gives you so little information that you'll never be able to tell what the app is, sorry about that. Currently I have a landing page up that describes the product, lists its features, and asks for an email newsletter signup from any prospective clients. I'm running a Google Ads campaign, with the help of a friend, that should help me to gauge the level of interest from my audience. Once that campaign is over, and I've collected all the data I need to move forward, I'll post a link to the landing page here so you can take a look at it as well.

In the meantime I'll be packing up my house and moving to a new zipcode. Hopefully, once we're settled in at the new place, I'll be able to pick this project back up and move forward with development. Only time will tell.

So, here I am in the middle of a month long sprint to finish a new application and what do I do? I jump ship for a couple days to build a throwaway site. On Wednesday last week I ran across an idea that was so funny, and so easy to do, that I had to push it out there so I could get on with my life! The outcome of that little delay is the new site 2things1page.com.

It's a really basic site, obviously, but the whole premise is just that you pick two things and it will create a page for you with pictures of those two things. In addition, if you click on the pictures you get a special surprise! So, like I said, it's a throwaway site that's completely for fun but at least it's out of my way now and I can get back to my current project.

All was not lost on my actual project though. Throughout the weekend I continued to sketch out designs for the main application areas and I plan to start doing some html mockups in the next day or two. Hopefully this little detour won't push my plans back too much.

Day 2 of my public product development schedule is now behind me and I'm making good progress. My goal for this point was to have an overview of the main application features and to write out what those features entail. So far I have about 5 pages of feature descriptions that break the application up into around 6 distinct sections. A good start!

My next goal is to do sketches of the main screens in the application to get a rough idea of what I'm building and how it will look to the user. I've always wanted a good excuse to use Balsamiq Mockups on a project and this looks like a great opportunity. Because of my tight schedule I need to really speed up the mockup phase and if Balsamiq lives up to its reputation then it should do the job nicely.

Maybe tomorrow I'll have a nice mockup to use for the image. As long as I can find one that doesn't give away trade secrets.

Recently I've been in a sort of personal rut. Normally I'm very active in various projects, I'm very entrepreneurial and constantly working, tweaking, coding, and learning with the goal of bootstrapping my own business into existence. Lately, however, I've really started to slow down. There are reasons for this, reasons that I'm sure any sane person would accept as perfectly legitimate, but my goal here isn't to be sane, it's to be Relentlessly Resourceful.

I decided that the month of March is going to be the month where I regain my entrepreneurial fire and really get something going. After reading, and re-reading, some great articles on startups over the last few days I've decided on a couple things that I hope will help me in this endeavor.

First, I'm going to live or die on this project publicly as suggested by Paul Graham in his essay 'How Not to Die'. The idea here is that if I set out to develop this project in the dark, as I usually do, then it's much more likely that it will die. Either I won't meet my, admittedly aggressive, development schedule or I'll half launch the app, mess with it for a couple weeks and then decide that it's not going to work and let it atrophy online somewhere. My hope is that by recording my development milestones on my blog I'll hold myself more accountable because I've publicly announced that I'm working on a new project and I expect it to make money. Anything less is a failure.

The second part of my plan is to constantly 'make things happen.' I have a very aggressive development schedule for this project, 31 days. My plan is that in by April 1st I will have a new web application deployed, in beta, that provides a valuable service to its customers and charges a subscription for its use. The current plan is to have a free option so that users can try before they buy, but I believe in the value of the software I'm building and as such I plan to be compensated for it. My goal is to build a business with my mind, hands, and computer. In a month. To make that happen I need to move swiftly, design a streamlined product that does one or two things really well, and implement it quickly, making progress every day.

So there you have it. Hopefully by posting my progress here over the next month I'll be able to fight my way through to a finished application, and the beginnings of a business, by April. If you've got any questions feel free to post them below, I'll try to be as transparent as I can with the development process. Hopefully anything that I can't reveal along the way will be available for you to look at in a month anyway, so you won't have to wait too long.

Picture Source

My shopping search engine Embought has just sent off it's Charity Donations to Child's Play! Embought donates 50% of all profits to charity and we are happy that this year we are donating to the Oklahoma Children's Hospital through Child's Play. We were able to donate some great baby toys as well as some fun books and movies for both younger and older children.

While it's obviously a small gesture I hope that these items can bring a little happiness to the children there as they fight their afflictions. Thanks to everyone who has purchased products through Embought and made this donation possible.

Merry Christmas and God Bless.

Last night the Boxee team unveiled the upcoming Boxee Beta release as well as the first ever Boxee Box. While the Boxee Box is pretty exciting in it's own right the Boxee Beta is what has me especially excited. It seems that with the Beta they have really gone out of their way to correct the biggest problems of the current version. Either of these two announcements would have been big news on their own but to get them both in one night was really a treat for Boxee Fans.

User Interface - Redefined

I've been a fan of Boxee ever since I started using its Alpha version on my AppleTV last year. It's a great Internet-centric media center that makes it easy to view media from your local network as well as tons of tv shows, movies, music, and podcasts from the Internet at large. While the idea behind Boxee has been great since the beginning the user interface and execution of the idea has been a bit lacking, until now. The Beta version of Boxee that was displayed last night looks fantastic, and it especially seems to have addressed the biggest User Interface problems that have plagued the current version.

Boxee - Now Working For You

The biggest problem with the current Boxee interface is that it requires the user to know where the content that they want is located on the Internet. To stream a TV show or movie a user must navigate to the correct application and then search for it within the app's interface. The Boxee Beta, thankfully, solves this problem by bringing all the media in your local library and on the Internet together and presenting it to you by type. So now you can browse and search Movies, Music, or TV Shows without worrying about what online site or service provides the file. This is a huge boost to the interface and will really help Boxee break out of it's current 'geeky media center' role and into a more mainstream entertainment solution. Another great feature is a personalized queue that allows you to tag shows for later viewing or subscribe to a show so that you are notified when a new episode is available, much like a DVR but without any of the hassle.

The Boxee Box

The other big announcement of the evening was the first look at the Boxee Box, and what a box it is. It seems that the team at D-Link took the unofficial name 'Boxee Box' and really ran with it as the design is described as a 'submerged cube'. It's a very striking design and it seems to have really polarized people online. I like the design, I think it's a bold move and one that will definitely stick out in your media center, but I can also understand why some people are put off by it. Other than the look the only details we have are the inclusion of Wifi, HDMI, 2 USB ports, Optical Audio Out, an SD card slot, and a $200 price point. There was also talk of using the USB ports for an external TV turner so it will be interesting to see if the software has any clues about possible Over the Air TV reception or DVR functionality.

I'm really excited about both the Beta and the new Box, I'm also un-patiently waiting for my invite email. I think that the new focus on user interface is going to go a long way toward making Boxee the software of choice among Media Center owners and hopefully CE and Connected TV makers. If you could get this software running on one of those new Vizio LED TVs it would be a dream come true. I'm actually so impressed and excited about the future of Boxee that I'm currently working on a Boxee App of my own. I'm hoping to have it available in time for the Boxee Beta general release in January so keep an eye on the blog for more info about that.

I'm an avid blog reader, although not quite avid enough at writing them, and over the last few days nearly everywhere I've gone I see stories about Apple and their mysterious 10 inch touchscreens.

I've definitely been burned by Apple Tablet rumors in the past but I'm still willing to get excited when this many different rumors and updates all fall in line together. Please understand that I have no inside information and that this post is all just hacker speculation, but if that is your kind of thing please read on.

A few months ago we started to hear rumors about a possible Apple Tablet, and this picture was photoshopped together as a reference. Since that time I've seen this picture on close to a million different blog posts. Since photoshop is not my forte I'll stick with it as a visual reference for this post, but it's important to note that the hardware isn't what is going to make this device worthy of Apple's time. It is, as always, about the software.

There have been quite a few people who have decried the Apple Tablet as a horrible idea, but they don't get what a Tablet device from Apple would actually be able to do. Just like the MP3 player market was nothing special before the iPod the recent batch of Tablet Computers, or MIDS, or UMPCs, haven't had any really interesting features to set them apart from a standard mobile computer. There is no reason to cram a fully featured OS like XP, or Vista, or OS X for that matter, onto a device that's made to operate on the go, with one hand. Apple knows this, and they've shown that to the world with the iPhone.

Another possible software solution for the MID space is the Moblin project. It is an attempt by Intel and various other industry partners to create the future of mobile computing with an open source Linux OS. Unfortunately, a few years after their 'concept video' was released they are still nowhere close to creating an easy to use mobile system, let alone a system as awe-inspiring as what's in the video. Check out the video below and tell me you don't get excited about the possibility. Then ask yourself if you actually see a group of industry partners and open source coders building that anytime soon.

I think more than anything, that video shows what an Apple Tablet is all about. The goal isn't OS X on the go, it's a hybrid system that runs a modified version of the iPhone OS on the go and then transitions you to a full OS X system when it's docked at your home or office. Apple has always made it's name by making things 'just work' and their best contribution to the tablet space would be this kind of seamlessly transitioning system. A device that gives you an easy to use interface when you're out in the world, and then gets down to business when you dock it at home.

This idea really started to gel for me when I heard the rumor that Apple is digitizing books for an eBook reader. I don't believe that Apple would just release an eBook reader like the Kindle, but I have no doubt that Apple would love to be the place students buy their eTextbooks when the next school year starts. An Apple Tablet like I've described that incorporated an eBook application would be the perfect device for college Freshmen next fall. Especially if it's released to replace the still plastic Macbook at the $999 price point. By eliminating the cheap end of the Macbook line and replacing it with this Tablet/Desktop device Apple could make it the must have item on every college campus next Fall, and begin to take a dent out of the $5.5 Billion textbook market with the iTunes Book Store.

Of course, this is all just speculation on my part. But if anyone can create a desirable Tablet/MID/UMPC it's Apple. In fact they may be the only company that can. Hopefully there will be some tidbits of knowledge to prove, or disprove, my theory at the iPhone OS 3.0 event next week. Until then I'll just have to watch that Moblin video two or three more times and imagine Apple logos on everything ;)

With all the talk about the economy lately I've spent much more time thinking about the stock market than I normally do. One day this week as the market seemed to plummet for no apparent reason I started to think about the overall dynamic of what makes the stock market work, all the people spread out all over the world valuing companies with their money. At that point something dawned on me, and perhaps it's something everyone already knows and I'm just late to the party, the stock market has been 'Web 2.0' far longer than the Web, so why did it take us so long to copy it?

Market 2.0?

Now I know that everyone hates the term 'Web 2.0', so do I, but just bear with me on this as I'm trying to get somewhere. One of the many definitions for Web 2.0 on Wikipedia says:

…the philosophy of mutually maximizing collective intelligence and added value for each participant by formalized and dynamic information sharing and creation.

The Market fits that definition almost perfectly, it maximizes collective intelligence, or fear and glee, to determine the value of companies. In rational times this value is crowdsourced based on the information provided by the companies themselves, factors effecting their market segment, or from respected analysts in the field. The end result, however, is always that the masses determine the companies value through their buying and selling prices, basically their votes. In essence the stock market is a giant version of Digg!

Other terms that have be thrown around in the Web 2.0 world like 'tipping point', and 'long tail' all seem to fit the market as well. The companies range from large in size with huge market caps and names that everyone knows all the way down to small companies that make kitty litter, and you can invest in them all. As for the Tipping Point, anyone who has looked at a graph of the Dow Jones over the last few months can definitely see that in effect. As stock prices started to slide more and more people started to sell their shares in fear. This tipping point created a market crash where there were almost no buyers because everyone felt they should sell.

So What?

So maybe I've convinced you that our Web 2.0 ideas are all visible in the Stock Market, so what? What is the point of this post? If you'll agree that many of the 'revolutions' of the Web 2.0 era were already visible in the Market for years it begs an interesting question. Why did it take us so long to see it? There are technological answers to that question of course, but surely we could have made a site that crowdsourced news headlines before Digg, so why did it happen just in the last few years when the market has been doing these types of things for much, much longer? Perhaps an even more interesting question is: What else are we missing? What other 'real-world' structures or institutions can we steal from to redefine the web? I'm sorry to say I don't know the answer right now, but it's interesting to think about, and I'd love to hear your thoughts on the subject too.

Embought is my first attempt at writing a search engine, and I really only started about 6 weeks ago. There is still a lot of work to do, especially when it comes to improving the relevance of search results, but I wanted to take a few minutes to chronicle some of my experiences over the last couple months.

If you have ever thought about writing your own search engine, to the point that you started looking around for information on the concept, you've probably seen this article on ACM Queue - "Why Writing Your Own Search Engines is Hard." This article is full of both great advice and scare tactics. In fact if you're not fully convinced that you want to write a search engine this article may very well scare you off from the endeavor, it originally did me. I first had the idea for Embought last fall and began searching for information on how to write my own search engine. I found the article on ACM Queue and quickly decided that maybe I didn't know what I was getting myself into.

In the article, after giving some great advice about how to avoid pitfalls, it's made clear that writing your own search engine is very, very hard and you should probably just use Nutch instead. Nutch is an open-source java search engine that sounds very impressive, and would probably have worked great for Embought, if I could make it work. After about 2 months of installs, configurations, debugging, etc. I never got a working system with Nutch. I never indexed a file, I never got a search result, nothing. The problem with using Nutch, in my opinion, is that it's built as a general purpose search engine system. It's like a giant Swiss Army knife with a million little blades. It's great that there are so many tools and blades on that thing but trying to use it for one simple task is an exercise in futility.

Now I don't want to make it sound like Nutch isn't a great piece of software, because I'm sure it is. What I'm trying to say is that Nutch was overly complex for what I needed to do. There are a lot of novel things that I do in the crawl and index stages of Embought that have made things much easier for me. To try and add that custom code to Nutch would have been a nightmare, I don't understand how Nutch works, and in the time it would have taken me to figure it out I was able to write Embought from scratch.

That's not a knife, this is a knife!

Embought is like a single blade knife, it has one purpose; to find products and present them to users. It doesn't need all those other 'features' because it's just a core codebase that does exactly what I want it to do. It's small, manageable, extensible, and fun to work with. It's everything that working with Nutch was not. I'm sure that Nutch is currently more powerful than Embought, but based on the feedback I receive from having a live system in 6 weeks I can rapidly improve Embought, I don't think I could have done with same with Nutch.

Steps and Tips - Crawling

After initially being scared by the ACM Queue article I later decided to use it as a guideline for my project. I built my own multi-stage crawler to scan the pages, just as the article recommends, and that has worked well. I have two additional tidbits about the crawling phase however, make sure you obey robots.txt from day one and make sure you have some amount of pause between multiple page requests on one host. I had a pretty nervous 30 minutes after I apparently crashed an online electronics retailer's website on a Friday afternoon. If any IT guys got called up to restart the server on that one I apologize.

Indexing

I'm currently crawling in phases of ever increasing depth. My latest crawl was of about 25 stores and 350,000 pages. Because I have limited resources in both the amount of bandwidth I can utilize and the number of available machines I've had to get clever about how I pull down the maximum amount of products for each crawl. I've started by writing a short block of code specific to each of the stores that I crawl. This code block is tailored to pull specific information from each product page into my indexer. For the actual indexing I'm currently using Ferret and I've had no problems. Ferret is fast, easy to setup and use, and so far very stable. I've heard that some people have corrupt index issues with Ferret but I've had absolutely no problems.

Hardware

As I said above I'm pretty limited when it comes to resources right now. I'm currently crawling on a linux pc and my G5 iMac. At index time I use those machines plus my PS3 running Yellow Dog Linux. The PS3 as a bare bones linux machine is pretty impressive, it indexes at more than twice the speed of my iMac, but it could definitely be faster. I'm currently writing some parallel indexing code that I hope will make the PS3 scream.

My future plans call for a much larger number of stores and products so I'm trying to be clever about how I scale my hardware. I'm hoping to test out linux on an Xbox 360 in the next couple weeks. The 360 has a triple core, 2 threads per core, processor that should be much easier to program for than the Cell. I'm also very interested in Nvidia's CUDA platform. If I can build a parallel indexing/ranking algorithm on CUDA it may help me to scale up my index tremendously, without much hardware cost. If Microsoft or Nvidia want to send me any hardware for this project I would be happy to sing their praises and post performance numbers.

That's all folks

I hope this has been somewhat interesting. I'm really excited about this project and I hope that it allows me, and all of you, to save money and give something back to the people who need it. If you have any insights of your own, or advice feel free to contact me.

Happy Shopping!

In case you couldn't tell from the fact that I recently wrote my own search engine, I'm pretty interested in search, I have been following the ongoing struggle between Yahoo and Microsoft pretty closely and this week has seen a number of big announcements. So here, whether you asked for it or not, is my two cents on the subject.

The Yahoo, Microsoft, Google, News Corp, AOL Soap Opera

The ongoing discussion between Yahoo and Microsoft has been well documented so I won't recap it all here. In recent days however numerous companies have had their names added to the fray in one form or another. Yahoo is now talking about buying AOL, and running a 'test' of Google Search Ads. Microsoft is joining forces with News Corp to institute a takeover bid Yahoo can live with, the drama goes on and on. The problem I have with this entire situation is that people seem to think Yahoo is doing all of this as part of some plan to ruin the well before Microsoft can take hold of it.

I think Yahoo is just doing whatever they can to persuade their shareholders that a Microsoft takeover is the worst possible thing that could happen to the Internet. Yahoo is close, really close, to being in a position to take on Google in search. Their search results are as good or better than Google's results. Seriously test out some searches, it's been my experience that you'll get the same results on Yahoo, or better, and that they will be more recently refreshed than what you'll find on Google. I replaced Google with Yahoo in my Firefox search bar a few months ago and I've had no reason to go back.

Why selling out to Microsoft is such a bad idea.

Google owns the search market, but that has the possibility to change. Yahoo is innovating in search with their Open Search Initiative, while Google is too busy trying to compete with Amazon on who can outsource their hardware better. The recent announcements from Yahoo that they will support microformats could really change the face of search on the web. By integrating semantic web information into standard search results users will be able to get a much better understanding of what the returned results actually mean. One example shown on the Yahoo Search Blog is a LinkedIn profile.

By pulling that information directly from LinkedIn and placing it into their search results Yahoo has created a means to keep people on their site instead of instantly clicking through, which means more visitor engagement and more ad revenue. Deep information retrieval and presentation is the next big thing in search and it's Yahoo, not Google, who is doing it right now.

The scary part of this entire situation is that most of the Yahoo shareholders don't know this, or don't understand it. They will file class action lawsuits against Yahoo for not selling out and put pressure on Yahoo to choose the absolute worst of their possible options. Microsoft will kill the innovation at Yahoo both by making the best and brightest flee the company, which they have already started to do, and by imposing their rigid structure on the Yahoo developers. There is no good outcome from a Microsoft-Yahoo merger, it was a mistake for Microsoft to offer it and it would be a colossal mistake for Yahoo to accept it.

What are the stakes?

A year from now the smoke will have cleared and we'll have one of two environments. If this merger happens we'll have a stumbling giant of a company in Microsoft-Yahoo facing a Google with even more search market share than they currently hold. Or we'll have a fight on our hands between Google and Yahoo for search dominance, a situation where all of us searchers stand to benefit. Search has been stagnant for too long, hopefully Yahoo can stay independent long enough to change that.

There has been a lot of talk online concerning the Single Founder Startup. Most of this talk is about how it doesn't work, you can't start a startup by yourself. I don't think that's true. You CAN start a startup by yourself but you need to be able to wear a few different hats to do it. It's the wearing different hats thing that gets most people, and why it is still a great idea to have multiple founders, but if you know what they are going in you might have a chance. I'm going to list off what these different hats are and why you need them in just a second but first let's get an assumption out of the way.

I'm writing this for the yCombinator crowd, which is made up of people focused on starting technical startups, usually online. If you're 'startup' isn't technical then I'm probably not going to be very interesting to you. You have been warned.

The 3 Faces of the Lone Founder

The Lone Startup founder needs to have 3 different faces; Tech Wizard, Artist, and Seller. The Lone founder also needs to be so driven that it seems the only thing he CAN do is start a startup. He must be so driven that he can't stop thinking about it. He can't stop working 10 hours a night on it after working 8 hours at his day job. He can't make himself go to sleep until he has working code that brings him one step closer to his product. The Lone founder cannot be a slacker in any sense of the word. In fact if you're a slacker, if you like it better when someone tells you what to do next, you don't need to be involved with a startup at all. You can't be a founder, and you shouldn't try to work for a startup because you're only going to be a drag on the other people that are sweating blood to make things happen.

So why does the 'Lone Founder' need those three specific hats? Because without any one of those three jobs your startup isn't going to make it. I know this can be debated but I'm just going to let you know where I stand on this. If you can't sell your product to a SPECIFIC segment of the market, or you can't make it comfortable and useable for them, it almost doesn't matter how great your tech is. It is so easy for someone to start a company now that if you only have good tech somebody is going to come along with a team that can copy your tech, make it easy to use, and sell it to the right people. You might as well be the one to make money off your technical wizardry so you need to get it right the first time.

So you need three faces or, if you're lucky enough to be blest with smart friends, three founders for your startup. Let's talk about what each one brings to the table.

The Tech Wizard

The Tech Wizard is 'the Woz'. He knows the best tool to use, the best way to make it sing, and the best way to get things done. This is probably what most of you are, or want to be, so I won't spend too much time here. If you know that you can make a good technical product then you're the tech wizard. If you don't like the idea of 'selling' to people or you can't make a 'pretty' UI for your tech then you need to look for somebody else to fill those roles. It doesn't have to be two people it could be just one other person, but you need both. To get started you have to have great tech, or at least implement something in a new way, but that is only the start. You also need:

The Artist

This one is probably going to create some discussion, which is good, but I believe that you need someone who can make things look pretty. I know there are sites that aren't pretty and they do just fine cough craigslist cough but if you really want to up your chances of people using your site it's better to make it pretty. However, the most important job of the artist is to make things useable, don't use ajax just because it's cool. Don't add a lot of fluff to your site for no reason. Simple can be pretty too, in fact I would argue that it is better to make your site simple with just a few visual cues that make it more enjoyable for the average user. You need to be able to make simple layouts that are constructed from the user's perspective instead of driven by the tech. Think about how people will user your site and design it around their needs. Real Artists make things useable.

If you can make great tech, and make it simply pretty and easy to use you're almost there. To be a real Lone Founder you need one more face:

The Seller

Notice I did not call this face 'The MBA' or 'The Business Guy'. There is nothing wrong with your Seller being a business guy or having an MBA but you don't want to look for somebody who just has those 'qualifications'. The seller is the guy who knows who the target market is. When someone asks the question; 'Who is your target market' he needs to know the answer. I'll give you a hint, the answer isn't 'Everybody with an Internet Connection', that is just who can use your product, not your target market. The target market is the 10%, or 5%, or 1% of the population who really NEED your product. The people who need it so bad they are willing to pay money for it. It's possible that your target market isn't even the people that will use your product daily. The fact that you have made something people want to use could be what creates your target market. [ Google gets paid through ad sales, so their target market is people who buy ads, while their users are 'Everybody with an Internet Connection.' See the difference? ] You need somebody to sell your product to your target market, and if that scares you or you don't know how to answer those questions you need to find somebody who can.

So there are the three faces, do you have them all? Maybe you do, if so stop reading this and get to work! If you don't have them all you have two options. The first is to find other people who can fill in the roles you can't do. But, what if you can't find somebody, what if it's just you? Then it's time to bite your thumb at the rest of the world and go it alone anyway. The best way to learn how to do all three of these things is to start failing as quickly as possible.

The sooner you get out there and actually build something the sooner you're going to fail. The sooner you fail the sooner you're going to have insight into what went wrong. The best way to learn how to do all the things you need to do in a startup is to start one. Hindsight probably isn't 20/20 in this case, but it's much better than your foresight. If you want to start a startup just do it. If it works out, good job, if it doesn't then at least you'll be better prepared next time.

Now go get to work!