Here is a link to my TiEcon presentation from today. I have posted the presentation with my speaker’s notes. In this presentation I introduce three ideas: Gr8, Personalizr, and Support.com.
Free Ideas. Just Add Execution.
May 17, 2008...1:13 am
My TiEcon 2008 Presentation
Jump to Comments
9 Comments
May 17, 2008 at 4:38 am
Hi Mike,
Thanks for posting the slides. Btw, it was easily the most interesting presentation of the day (in my opinion).
After hearing from you, I have got a related idea - outsource development of algorithm to the end users. This way the users will feel involved and will definitely give what they are interested in. Less risk of creating a product which users don’t want :-). In this era of web world, A/B test will determine which algo is “better”.
Thanks,
Ankur
May 17, 2008 at 3:13 pm
Thanks Ankur. Opening algorithm development to any third party was the core idea behind “Gr8″ in the presentation.
May 17, 2008 at 10:50 pm
Oh no, now my “personalize the web” idea is out in the open
Many of your points resonated with me - especially the “more time doing, less time protecting.” One of the reasons I invested and joined the board of a start-up called Path101 was their founders took the “anti-stealth” approach - putting meeting minutes, investor pitches, etc online for feedback. More disclosure and sharing will speed innovation cycles.
As for “if people think you’re crazy, you might be on to something.” I was the first non-engineer at Second Life back in 2001. That experience helped me understand that there are two types of entrepreneurism - opportunistic and evangelical. Opportunistic is the innovation against a path or outcome that’s widely understood to be accurate but you create value by getting there faster, better, etc. Evangelical is when you see something that others don’t - and they think you’re crazy - and you just might be. But your job every day is to get more hands on the boulder to help push it up the hill before it rolls back over you. Evangelical entrepreneurism is a blast but it also can be really hard on founders because you need to balance self-confidence with being able to process contrarian data effectively. And because, well, you just actually might be crazy.
May 17, 2008 at 10:56 pm
Thanks Hunter. Love your comments on the crazy versus brilliant bit. You are absolutely right that sometimes people just are crazy.
A friend of mine was recruited by Google in the late 1990’s. He came out of a meeting with Google and was shocked. He was a Phd. in Mechanical Engineering from Stanford and the Google guys wanted him to build robots to grab books off a shelf, take them apart, scan them, and then put them together and put them back on the shelf. They wanted all of the world’s information to be available and they didn’t let conventional wisdom get in their way. Years later they did just that, but they used low cost Indian labor instead of robots. At the time he thought they were crazy. I agreed with him. We were both clearly very wrong ;–)
June 1, 2008 at 11:10 pm
[...] article was an overview of the recent TiECon 2008 event that held in Santa Clara, CA. One of the key takeaways was that economic downturns are [...]
June 2, 2008 at 4:10 am
[...] recently stumbled upon a presentation by Mike Speiser (on his blog) which promotes the notion that good ideas should be shared. He writes: Sharing your idea will [...]
June 25, 2008 at 5:55 pm
Just stumbled across these slides - thanks for sharing! I wasn’t at the actual talk, so apologies if I misunderstood something from reading the slides
I’m skeptical of your first idea. I have some experience in data mining & machine learning algorithms, and from my experience the quality of the data matters much more than the algorithm. Great data with crappy algorithms beats crappy data with amazing algorithms anyday. The reason Monster, Google News and ad targeting are terrible is not that we don’t have good algorithms to target it to you. It’s that we (the people developing the algorithms) don’t have enough data about you. If I knew your age, profession, gender, job, interests, etc, even a very crappy algorithm would produce decent results. From my experience, it’s all about the inputs.
Opening up an algorithm platform to 3rd parties might provide a 10-20% improvement in quality. But you’ll see the huge jumps in quality only when Monster.com “knows” your industry, current job, jobs you applied for, your friends’ jobs, etc. Or when Google News “knows” the 3 most interesting articles you read each day for the past 3 months (right now Google News is missing a rating system — all it has is your click data).
There’s only so much you can improve with really limited data.
July 14, 2008 at 4:06 am
Here are the existing sites that solve your problems:
A) Some algorithm that can pretty much solve any personalized relevance problem. Whoa. No answer there… perhaps Google?
B) Taking away registration- Check out clickpass.com, the world of OpenID is what you’re wanting.
C) One stop support- lots of sites trying to do this. Perhaps the best so far is Get Satisfaction in which you can see how a representative fixed the previous problem you talked about, so at least you have an answer. Additionally, you allow your customers to help answer your questions too.
July 18, 2008 at 6:11 am
I just came across http://getsatisfaction.com/ today, and I’m going to second Andrew’s recomendation.
This is not quite the Support.com you envisioned in free idea #3, but it’s the closest thing I’ve seen yet — I’m a fan.
Leave a Reply