If you believe what Github says, then this piece of software took 9 years and 3K commits to develop but that's leaving out a couple of years of analysis and probably a couple of thousands of commits done in the dark. And yet I'm only half-way done. But not to worry, the final pieces are easy: (1) implement a semantic language model, (2) implement sharding to enable you to distribute your data across many machines and (3) implement voice-to-text. That would put us pretty darn close to us having a out-of-the-box turn-key Google-killer FOSS solution for anyone to deploy to the internets and make billion dollars on.
Which brings us to my proposal, which is, don't pick Lucene or Elasticsearch when choosing your Google-slaying tech. Instead, pick a piece of tech that is specialized on solving the problems you'll face in a web search engine context, that is maintained by someone who tirelessly try to outdo all of the state-of-the-art FOSS search tech out there, even though he is a single-member team.
1 comment
[ 3.0 ms ] story [ 14.7 ms ] threadWhich brings us to my proposal, which is, don't pick Lucene or Elasticsearch when choosing your Google-slaying tech. Instead, pick a piece of tech that is specialized on solving the problems you'll face in a web search engine context, that is maintained by someone who tirelessly try to outdo all of the state-of-the-art FOSS search tech out there, even though he is a single-member team.
Also, feel free to help out.