So can someone tell me...
 

how long does LW search for? Im sitting here watching it and after about a couple of minutes the green bar in the search tab stops growing and my search results stop. That cant be it, can it? Shouldn't it actively search until i click stop? I do really enjoy using this software, but ive been curious about this for a while now because sometimes i get very limited search results.

Searching indefinetly is bad for gnutella, believe me if it was the case, ultrapeers would be so overloaded that you would have difficulties to get a fraction of the hits you get now. LW tries to find 40 single results/ultrapeer you connect to it (ususally you connect to 4 (40*4 =160) or 6 for the pro version (40*6 =240) that's it. There is another way to search more (à la shareaza) but this method is unbelievely slow and is killing their own network (UP overloaded).

LW searches for what--4 or 5 hops that return results?--before the query is dropped, and this means about--what? 50%? of the current network. However, don't more specific queries effectively search further?

Yeah when a LW UP doesn't find 40 single results (not counting alternate sources coming from a single host) it can send a search with a TTL of 4 (even if they advertise the max TTL header of 3). I guess you can reach 30-40% of the high outdegree and dynamic querying network clients (Bearshare + LW) which probably means 100 000 hosts.

i see, so longer search queries would actually harm the network? Well that sucks. I guess the network isnt really big enough yet to handle a feature like that.

ANY network no matter the size can't substain that thefuzmixman. Just think about Kazaa lite, it litterally destroyed the fasttrack network with his automatic search more (one could argue that RIAA and the spyware in kazaa destroyed fasttrack, but kazaa lite has a HUGE responsability). Another exemple, Shareaza with is "supposed" global searching has only made searches go much slower on G2, because popular searches are bringing it down to it's knees like it happened two years ago on gnutella (thousands of results for a search like .mp3). Less = better sometimes especially in decentralised P2P of today.

K-lite killed fasttrack? I never really thought about it like that. Im confused, as i always thought ft died out from massive riaa ploys and well... fake files being spread out in abundance, viruses etc... not broader search queries.

Nobody was really complaining for the huge ressources taken by Kazaa lite because Kazaa was a closed protocol (gnutella is open and many people check the traffic for irregularities). Also KL was liked by the P2P apologists while Sharman (new owner of Kazaa) was hated because of many spywares. The creators of Kazaa (the protocol core still in use) haven't touch to the protocol since Kazaa 1.5 and they are pure geniuses: they made Kazaa into a performant protocol (in 1 year) while gnutella only reached and outpaced Kazaa since fall of last year (3 1/2 years in the making). Now they are working on Skype another internet jewelry. Did you know that in kazaa 1.5 they limited the number of results to only 50, and only later they let users get max 200 results? They did this on purpose, their protocol, routed by big ultranodes (users very fast connections with up to 10 000 leafs) was limited by the numbers of searches routed. At some point too much searches were performed, and the performance decreased a lot for all the network.

Since Sharman doesn't have a clue about the fasttrack protocol, nothing effective was done and fasttrack is dying ;) But others will take it's place, c'est la vie.

http://www.skype.com/

^^ wow. that actually seems pretty cool. maybe not what i was expecting when you informed me that the makers of kazaa had a new project, but still can be useful. Anyways, i guess you've justified limewire's reason to have limited searches pretty well, and now i stand more educated in the matter... :p