Hey guys, thanks for your comments, both critical and supportive. To start I'd like to thank everyone so much for their support during the Beta phase. Without it I might have given up. I would like to thank the critics as well, because criticism is a powerful motivator. I want my tool to be perfect, and criticism both points out the faults and also drives me to make the critic happy.
Sorry I haven't responded earlier- I've been too busy developing the next version of EstiBot
The criticism here is mostly valid. Right now things are especially bad as Overture is down. I've built a contingency to EstiBot that when Overture is not responding, it will use a different algorithm, but the truth is that if you don't know the search popularity, it will be impossible to get a very good appraisal.
I am actually quite surprised that the current appraisals are not as bad as I feared, even without Overture data. The algorithm has become exceedingly complex, and any change that I make takes a lot of time because I have to consider the impact of even a small change to the algorithm as a whole.
In a way am happy that OVT is down, because it has got my ass in gear to develop the next version. I'm happy to see that it does a decent job even without Overture, which to me means that once I get the search popularity issue sorted out, it will be good. And there's more in the works.
The Beta of this EstiBot is approaching its end, and I am grateful to all users for their feedback. I've learned a lot, and hopefully the tool has been of some use to some fellow domainers.
If OVT comes alive again, I will adjust the current version to again consider the OVT and you should be getting better appraisals again.
Many users complain of instability; this is true, and while some of it is due to my programming (which I have mostly been able to fix), mostly it is due to the various resources producing unreliable results. EstiBot relies on OVT, google, yahoo, alexa, and even their API services are not reliable.
I alone know how the algo works, and I've tested it extensively, and I am still entirely convinced that the principle of my algorithm is solid, and that, given the right resources, I can and will produce an autoappraiser that will convince even the nonbelievers.
The current version is not nearly as good as I hoped it would be. More than 50% of the 2,000-odd lines of code is dedicated to overcoming the shortcomings of Overture. About 20% is dedicated to overcoming the caveats of the other resources.
This is why I have now decided that I will not bring this EstiBot out of beta at all. I will no longer make changes or improve it. It's the final version now. The beta has shed so much light on what a truly good auto-appraiser should be.
I'm pleased to say that EstiBot v2 is now in the making. It will utilize a different approach. The core principle is solid and will stay the same, but the algorithm will be hugely improved thanks in big part to collaboration with Justin's NameBio.com, as well as some other resources, and new findings on what contributes to domain value.
I am now putting everything I've learned during the beta to use in making a brand new EstiBot. I am doing all that I can to get access to such powerful resources that will really enable me to bring my appraisal principle to its fruition.
The principle is this: each metric has to be proven to affect domain sales value to be included in the domain valuation. The effect of each metric, alone and in combination with other metrics, is analyzed by complex mathematical analyses. This is then correlated to known past sales. This is the heart of the EstiBot system, and as I find more such metrics, I add them to the valuation, and the result is increasingly accurate valuations.
You can't call domain appraisal a science, but scientific method can be harnessed to make a good domain appraisal system. I have put my scientific training to use in finding what really, objectively, makes a domain name valueable. In my analyses, I use the same strict scientific criteria that are used in scientific, for instance medical, research. The current version is rudimentary; yet in many people's minds it is already the best autoappraiser out there.
One benefit of this method is that it reveals, which factors actually contribute to domain value and which do not. For instance, many think that domain name length is a factor, when in fact it is not, except in subgroups of domains that consist of only 2 or 3 characters, and on the other hand at the other extreme end of very long domains. But in the vast majority of domains, length does not correlate with value no matter how you test it.
Another benefit is objectivity; bad calls are made all the time by human domainers because they have a subjective view on the value of their domains. Objectivity, however, is also a weakness: An autoappraiser will never be able to match the expert human mind because there are so many psychodynamic factors at play in the domain market. However, most of us don't have an expert human mind handy at all times- that's why a good objective appraisal system can help.
This post is getting very long, thanks to anyone who has bothered to read this far
Hopefully, soon I will be able to present to you a tool that you will find satisfactory.
Some features that you can expect:
-lightning fast appraisals
-extensive keyword report
-bulk feature, up to 100 domains at a time
The current test version of EstiBot v2 analyses 100 domains in 40 seconds.
Thanks!
Josh
