SKA news

[Fungus]

Software licensing seems rather attractive option - having good show case of it in terms of huge WWW database is the kind of demonstration that will make it sell far easier than otherwise - because I have to scale to far far bigger volumes than ordinary SE writers my stuff will be much more efficient with small volumes measured in millions of products. That's serious competitive advantage.

Software licensing to who? And why?

Mike Lynch, if I recall the name corerctly, is doing that with his search engine technology. Can't remember the name of his company. Something like Infinity.

[AtW]

Software licensing to who? And why?

To companies that have search engines - specifically etailers as they have lots of product lines and home brew engines are often taking lots more resources than necessary, I know that since I rewritten search engine for my ex-ex-employer - that one was done by expensive consultancy and no wonder it was crap.

Mike Lynch, if I recall the name corerctly, is doing that with his search engine technology. Can't remember the name of his company. Something like Infinity.

The name you should know is Matt Wells - founder of Gigablast. He is pretty much the only man who single handedly reached big scale for his search engine and he licensed it to many startups. Last year he offered 25% of shares in his company for multi-million dollars payout. He did it in 4 years and I am fairly close to reach his level (technology wise) in just over 1 year.

And who else can say that? You see to do what I do is not easy to say the least, that's why you don't see many search engines with multi-billion page indices.

[bogeyman]

You see to do what I do is not easy to say the least

True, Bro - True!

[Fungus]

Hear, Hear!

Well said Mr. Fungus (you're not a Bogeyman perchance?)

Here's some snaps of my family...

Please, I've just had my dinner. That second picture is disgusting.

[Fungus]

To companies that have search engines - specifically etailers as they have lots of product lines and home brew engines are often taking lots more resources than necessary, I know that since I rewritten search engine for my ex-ex-employer - that one was done by expensive consultancy and no wonder it was crap.



The name you should know is Matt Wells - founder of Gigablast. He is pretty much the only man who single handedly reached big scale for his search engine and he licensed it to many startups. Last year he offered 25% of shares in his company for multi-million dollars payout. He did it in 4 years and I am fairly close to reach his level (technology wise) in just over 1 year.

And who else can say that? You see to do what I do is not easy to say the least, that's why you don't see many search engines with multi-billion page indices.

The proof of the pudding is in the eating. But they do say that pride comes before a fall.

I can't see why an etailer would pay for your technology unless it was cheap. Most don't need speed. Who on earth would need multi-billion page indices? Most etailer sites are tulipe anyway! But if you can succeed, then good for you. But I remain sceptical.

It was indeed Mike Lynch:

http://www.telegraph.co.uk/money/mai...9/ccprof29.xml

He has a history of successful companies. But I would never trust a man with such a neat beard. It's far too creepy.

[Churchill]

**unacceptable. please restrict that language to 'light relief'**

Oh, you're one of the BIG BOYS here then are you?

Sigh!

Oh Dear, are we suffering from a restricted vocabulary?

[AtW]

I can't see why an etailer would pay for your technology unless it was cheap. Most don't need speed. Who on earth would need multi-billion page indices?

I want multi-billion page index, however if I achieve high performance for that number it means that my stuff will be super fast for much smaller indices - like multi-million page one.

Why would etailer need faster searches? I explained it but you not listening - much faster search means needing less hardware! Can't you get it? Instead of paying rent or buying 4-cpu box one can get away with just 1 CPU!

Add to this increased relevance of searches and if you look at internal cost of developing in house search engine the benefits are pretty clear.

Having a multi-billion page index is a great showcase of technology and also suits my ultimate purpose of having WWW search engine.

[Fungus]

I want multi-billion page index, however if I achieve high performance for that number it means that my stuff will be super fast for much smaller indices - like multi-million page one.

Why would etailer need faster searches? I explained it but you not listening - much faster search means needing less hardware! Can't you get it? Instead of paying rent or buying 4-cpu box one can get away with just 1 CPU!

Add to this increased relevance of searches and if you look at internal cost of developing in house search engine the benefits are pretty clear.

Having a multi-billion page index is a great showcase of technology and also suits my ultimate purpose of having WWW search engine.

Yes but which etailers would need such powerful software? Very few, apart from Amazon and other biggies. In which case you have a small market. Most would run fine on one CPU. In fact I presume that for most etailers it is not the search that imposes the load, but the rest of the kaboodle.

Have you talked with etailers to find out what they want?

Fungus

[AtW]

Yes but which etailers would need such powerful software?

Search engine implementations are notoriously ineffective - in my ex-ex-employer we had search engine for 250,000 product lines it was slow despite running on 12 (or 8?) CPU Sun box with I think 12 GB of RAM, after I rewritten it overall database load dropped from 80% to 25%. This gave chance to get rid of very expensive box and replace it with much cheaper one, enough savings for all the salary I received in the years working there, sadly the company was totally %%%%ed so that such improvements were drop in an ocean.

Thus, there are big savings in terms of hardware, and also improved relevance - if people can't find stuff easily on a site they run more searches - this means slower - or just go away.

[bogeyman]

Search engine implementations are notoriously ineffective - in my ex-ex-employer we had search engine for 250,000 product lines it was slow despite running on 12 (or 8?) CPU Sun box with I think 12 GB of RAM, after I rewritten it overall database load dropped from 80% to 25%.

250,000 product lines? 12 CPUs? 12Gb? Search slow?

250K lines is a mere trifle! Should have been quick on even modest hardware.

What the hell did people search on? Must have been very complex queries full of disjuncts.

Wow! Now I realise I'm dealing with a super-dooper search expert

Can't find emoticon for ****** so banana will have to do.