4 ms·

I agree, and I think the author does too. Note that they said they were "putting it back on the shelf" and were willing to come back to it. I think distance an

I agree, and I think the author does too. Note that they said they were "putting it back on the shelf" and were willing to come back to it. I think distance and time can help with perspective, too. In a number of projects, stepping away (especially during the rat-holing on useless cruft features or w/e sections) and coming back later can help bring some good clarity.

xmprt 3y ago
Author here. I think all the buzz that this is getting make me want to give it another shot. I'll probably throw away most of the code and start from scratch though - one thing I learned from programmers much smarter than me is that the first solution is never the best one.
- fuzzfactor 3y ago
  Can't read the artice but I do have some generic observations. Most projects aren't going to work, so that can be one of the primary goals, mainly to find out whether or not that's the case, without too many compromises. So most of the time to be realistic you have to try real hard for it not to work quite early on, or that could be something where you end up wasting too much time on a hopeless effort. Some people never quit, and it can be difficult to know when to mothball something or give up altogether. Mothballs can be your friend, especially when there are a greater number of promising projects than could ever be fully completed anyway. Progress can surprise you though, but more often than not many things will take an unexpected downturn before you ever see a case of an unexpected upturn.
- janalsncm 3y ago
  I don’t think you need to throw your code away, I think you need to fix your ranking algorithm. People use Craigslist even though it’s got minimal styling. CSS isn’t UX. And to fix your ranking algo you need to understand places more deeply. The core of your product. You mentioned yourself that ranking by page views was flawed. First of all, it seems that you can probably categorize places into a distinct set of categories. Historical landmarks, locations of past events, locations of ongoing or regular events, buildings for large businesses or government, metro areas (cities, counties, states) etc. Figure out what you actually think is interesting. Probably don’t include a link to Manhattan on your map next to a local business for example. Choose a city you’re not familiar with and figure out if the top links are interesting to you. Tweak your algorithm and check another city. Simple isn’t always good, the world is messy and you’ll have a lot of edge cases.
- jll29 3y ago
  Think about the _user_: Is your use case "I'm a tourist in New York, who would like to learn more about my whereabouts." or "I live in New York, and I would like to be surprised with historic facts about places in my community", or what?
- hyperthesis 3y ago
  Divide into testing market/user-experience vs technical hypotheses: "Would it be fun" you can test by faking it, like curating it by hand for a small specific area (e.g. a popular street in a popular part of a popular city; or your hometown; or somewhere that appeals/you want to visit). "Can I make a general algorithm using publically available data to do this for everyone, for everywhere, automatically" is the technical hypothesis. It seems like a really challenging problem! But because it is clearly defined and scope-limited, I feel it's more straightforward than the first one.
nonrandomstring 3y ago
Yeah "the shelf" is where projects go. Never give up. There's some kinda weird zombie projects though; Project Deja-Vu: halfway through a new project you have a vague recollection of having done something similar before, and when you search your archives and find exactly what you're working on - even with the same comments (maybe this is an age related thing ) Project "Father give me legs...": Every time you look at it you're overwhelmed with a wave of guilt, memories of broken promises... Project impossible: It's a brilliant idea, everything works except that one NP-complete quantum-entangled function to send SMS text messages backwards in time to remind yourself not to do something.

xmprt 3y ago

Author here. I think all the buzz that this is getting make me want to give it another shot. I'll probably throw away most of the code and start from scratch though - one thing I learned from programmers much smarter than me is that the first solution is never the best one.

fuzzfactor 3y ago
Can't read the artice but I do have some generic observations. Most projects aren't going to work, so that can be one of the primary goals, mainly to find out whether or not that's the case, without too many compromises. So most of the time to be realistic you have to try real hard for it not to work quite early on, or that could be something where you end up wasting too much time on a hopeless effort. Some people never quit, and it can be difficult to know when to mothball something or give up altogether. Mothballs can be your friend, especially when there are a greater number of promising projects than could ever be fully completed anyway. Progress can surprise you though, but more often than not many things will take an unexpected downturn before you ever see a case of an unexpected upturn.
janalsncm 3y ago
I don’t think you need to throw your code away, I think you need to fix your ranking algorithm. People use Craigslist even though it’s got minimal styling. CSS isn’t UX. And to fix your ranking algo you need to understand places more deeply. The core of your product. You mentioned yourself that ranking by page views was flawed. First of all, it seems that you can probably categorize places into a distinct set of categories. Historical landmarks, locations of past events, locations of ongoing or regular events, buildings for large businesses or government, metro areas (cities, counties, states) etc. Figure out what you actually think is interesting. Probably don’t include a link to Manhattan on your map next to a local business for example. Choose a city you’re not familiar with and figure out if the top links are interesting to you. Tweak your algorithm and check another city. Simple isn’t always good, the world is messy and you’ll have a lot of edge cases.
jll29 3y ago
Think about the _user_: Is your use case "I'm a tourist in New York, who would like to learn more about my whereabouts." or "I live in New York, and I would like to be surprised with historic facts about places in my community", or what?
hyperthesis 3y ago
Divide into testing market/user-experience vs technical hypotheses: "Would it be fun" you can test by faking it, like curating it by hand for a small specific area (e.g. a popular street in a popular part of a popular city; or your hometown; or somewhere that appeals/you want to visit). "Can I make a general algorithm using publically available data to do this for everyone, for everywhere, automatically" is the technical hypothesis. It seems like a really challenging problem! But because it is clearly defined and scope-limited, I feel it's more straightforward than the first one.

nonrandomstring 3y ago

Yeah "the shelf" is where projects go. Never give up. There's some kinda weird zombie projects though; Project Deja-Vu: halfway through a new project you have a vague recollection of having done something similar before, and when you search your archives and find exactly what you're working on - even with the same comments (maybe this is an age related thing ) Project "Father give me legs...": Every time you look at it you're overwhelmed with a wave of guilt, memories of broken promises... Project impossible: It's a brilliant idea, everything works except that one NP-complete quantum-entangled function to send SMS text messages backwards in time to remind yourself not to do something.