I hope I don't come across as negative here, but I
hate lists like this.
Peopel are always raining on each others' ideas
"Oh, what, you're going to make a myspace clone? Bahahahaha." (facebook)
"Oh, what, you're going to make a facebook clone? Bahahahaha."
(twitter)
"Oh, what, you're going to make another dating website? Bahahahaha."
(Okcupid, plentyoffish)
"Oh, what, you're going to make another social bookmarking website? Bahahahaha."
(Reddit, digg, hacker news, etc.)
This sort of "You can't improve on an existing design" rhetoric is usually coming from the same people who champion companies like netflix and say things like "The RIAA suing downloaders is like wagon wheel manufacturers suing car tire manufacturers! GET WITH THE TIMES, HELLO!"
You want to make a better dating website? Awesome, I hope you do and I hope it's better than okcupid! If it fails, guess who's going to come out the other side better than they were when they started?
You want to make an abstract machine learning system? Good! I hope you do! Guess who's probably going to be learning a lot about ML when they come out the other side?
You want to make a craigslist/ebay mashup? Good! Do it today! Start it right now and, if it's better than craigslist, my friends and I are all going to use it! How much are you going to learn about interface/UX design in the process?
You want to apply gaming mechanics to exercise? Good! Do it and tell me about it! That sounds awesome! Make an iPhone app for it, make a facebook app for it, let me pick random strangers across the internet to challenge at it. Make leaderboards and lots of badgers, and blog about it. If you fail, write more blog posts about why you failed.
I'm sorry, I'm sure the author of this blog post had good intentions; trying to help other geeks, but I find this sort of "don't even try this because it's a stupid idea and you're going to fail" attitude extremely harmful.
The first project I ever did was called http://newslily.com it was a fark/reddit/hn/digg clone. Did I sell it to google for 2 billion dollars and retire to a yacht somewhere just east of anywhere on the planet? No. Did I go from knowing absolutely nothing whatsoever about web development to being able to turn ideas into things? Yes.
My current project is called http://thingist.com . Have people said "oh, psh, you're making a twitter clone...boorrrrriinnnggggggg."?
Guess who doesn't care? My daydreaming about having 50 millions users is forcing me to learn about scaling, and how to use mod_python (oh, and if I get super crazy, maybe nginx as a web cache for my 10 users!). In a year, or two months, or six weeks, or however long it takes me to decide that, yeah, well, it's just another twitter clone and isn't going to get more than 10 users, I'm going to be 1 twitter clone closer experience wise to making something people love. (Although I still think that thingist isn't a twitter clone).
My advice: make 10 twitter clones and 20 abstracted ML frameworks. NOT doing this is like a running coach advising their runners not to waste their time jogging around their neighborhood because they're really not going anywhere anyway.