Show HN: A Twitter crawler for app ideas | Hacker News Reader

Show HN: A Twitter crawler for app ideas

iwishtherewas.app

272 pointsmulhoon124 comments

> I wish there was an app or website that steals my tweets and aggregates ideas on the site.

From https://twitter.com/httpsterio/status/1380313273917128712

Many of the tweets of the website right now are meta-tweets referring to HN, the website or similar.

This is great. Many years ago (5 or 6?) I started pulling down any tweet that had the phrase “there should be an app...” and saving to a database daily with the idea of making something similar to this. The script is still running and I have about 40,000 tweets saved. I think I’ll pick it back up, thanks for the inspiration!

For MVP you could just have a bot account that retweets!

I could see these 2 bots re tweeting eachother

There should be an app that retweets "there should be an app" app ideas.

I wonder if thats enough to train a good idea generator AI

Idea generator? Yes. Good? Hmm...

Maybe fit it into the GPT3 engine and see what brilliant idea pop out.

Open source it

Wow this is fascinating just for the entertainment. Especially liked this one...

> I need an app that sends me a notification 90 seconds before the sun reaches the exact position in the sky where it reflects off my neighbor's windows and blinds me.

https://twitter.com/DanEngler/status/1379616188020318208

I feel like Shortcuts automation in iOS could do this.

Working on it... [1]

[1] https://shademap.app

Whoa that’s really cool! Also are you using some kind of IP to location? Was impressed you got my city right without needing location access

Edit: also super impressed by how fast the “shadow layers” are calculated. I’m in Switzerland and people here will love this ... knowing exactly when you’ll get sun in a valley is super useful. This might be something you could sell to people like search.ch - they could combine with the weather at meteo.search.ch - if you want an introduction say the word

I do use an IP to geo database. It puts you in NYC if the IP is unrecognized.

I checked out search.ch. Looks like a great add to Weather Animations section. I’ll try reaching out to you via Twitter. In the meantime please share with anyone who finds it useful. I use it for hiking and snow conditions myself!

"I wish there was an app like find my iPhone but “find my glasses” because fuckers keep getting lost"

Argh, can relate. Terrible catch-22 when you can't see well enough to find your glasses. Also, for reasons I don't understand, my wife owns a zillion pairs of cheap readers that sort of look like my glasses.

My version of this app would start with "what's that on the top of your head"?

You'll have to go a little further back. After I switched to round lenses, I couldn't see the frame as well in my peripheral vision, which led to me looking for my glasses as they were on my face.

Hold your phone up near your face with the camera on. You should be able to see the room clearly.

You could try Tile. Should be possible to attach one to a glasses necklace.

My favorite:

hayo @red_hayonnaise

I want an app that takes me on a timed, random drive; like I wanna drive around for an hour, map me a course that winds backroads and brings me back home in that timeframe

This would be spectacular if it could find fun/twisty roads nearby, similar to https://www.motorcycleroads.com.

I've used Twistypedia to learn about such roads https://www.twistypedia.com/

Interesting.

I wanted a similar app, but A to B, rather than round trips.

Best I could do with the majors is disable highways & toll roads, but often it would still try to take similar routes.

BJ Landcruiser tops out at ~120k, and who wants to go in straight lines on highways. More fun to try and find a fire road it can’t get down. Still searching.

Maybe the custom routing from GraphHopper is what you want: https://www.graphhopper.com/blog/2020/05/31/examples-for-cus...

You can eg. avoid primary roads additionally to motorways and tolls.

(note, I'm a co-founder of GraphHopper)

Not only for car but also for motorcycle, bike, walking, ...

Can be further tweaked with preferring/avoiding curvy roads or elevation.

And GraphHopper also supports round trips.

OsmAnd~ might have this capability. It has ways to avoid main roads when I use it cycling.

Very cool product.

I'd probably take the mountain biking directions, until the proper backroads and then hema maps are the best so far.

https://graphhopper.com/maps/?point=Gisborne%2C%203437%2C%20...

Lets me know if you ever do 4x4 mapping, although I can't imagine there's a strong business case for it.

We have a simple 4 wheeler config: https://github.com/graphhopper/graphhopper/blob/master/core/...

It’s for running rather than driving, but Trail Router finds round trip routes:

https://trailrouter.com

HN discussion here:

https://news.ycombinator.com/item?id=23802317

My GPS cycle computer does something similar - plots a random route close to a fixed length.

What model cycling computer does that?

Garmin Edge Explorer 820 has onboard route planning and I have the whole of Europe on the device! Amazing little unit.

An idea with no societal downside.

We definitely need ethics to be a core part of software dev training.

Well, I guess you'll have to weigh the amount of fossil fuels that you burn doing that against whatever else you were going to do in that time.

Unsure if that's a joke, but no! That is absolutely not what you would rely on! You can't simply delegate all ethics to the level of the individual user! Your actions matter, the technology you create shapes the world, and you have to consider the consequences of it.

I was eyeing up a satnav for my motorcycle, one option it had was to create a windy, indirect, touristy route instead of the fastest / most efficient one. It was an interesting one.

But in the end I couldn't justify paying €400 for something that my phone could do. I know, motorcycle satnavs are hardened, deal with direct sunlight better, work offline and are much more energy efficient, but still. I feel like at that time (this was only a few years ago) those systems should be a lot better already, like offloading work to your phone (becoming more of an external screen), etc.

Just set an alarm, turn off all your devices and get lost. When the alarm goes off, reconnect and go home.

We've lost so much magic in our lives because the phone is too easy to use. We've eliminated serendipity

"What the fuck do I need to do to stop seeing feet on my timeline why do people keep posting them? Is there an app that blocks all feet? Send help not feet"

helpless giggling

The user can feedback the algorithm with "show less like this" or block the author accounts.

Ironic, just before I read this comment I'd visited the crawler and the top teeth there was a foot fetish (the username mentioned feet and an onlyfans) twitter account wishing for an app to track who engagued with their tweets to run a contest or something.

Lots of interesting ideas on here! Very cool site.

Here's a pair of related tweets I found:

> dear coders: I need an app that will scan receipts from contractors, automatically pull the relevant info re: business name, type of service, date, receipt #, etc, and file it in my phone contacts with a 1 year reminder on my calendar for future maintenance calls.

https://twitter.com/LDBurnett/status/1379785147294310405

> I need an app that’ll remind me when my cards are about to expire. Cards, passport, even products. Need.

https://twitter.com/newii/status/1379439739615514627

Together these might form an interesting app idea.

Maybe I'm old school, but this seems like something that would take upwards of five minutes to add to my calendar and wouldn't require a whole other account + login.

Yeah I find that calendar and a free-form note app cover pretty much everything "miscellaneous" on my phone. The only other apps I really use are maps, contacts, email, and web browser.

Okay I have a legitimate question: for random things you need to do in a long timeframe (ex. Change the Air Filters in 6 months) do you add it to your calendar? Siri for example would add it for exactly 6 months to the day from now, and what if that is a Tuesday and you'd rather do it at a more convenient time?

Adding it to reminders would have an enormous number of pending reminders it feels like

As soon as it pops up on your calendar, it's now a 'to-do'. Manage that the same way you manage your other to-dos, wether that's a reminder or a sticky note or a list or whatever.

I would see myself procrastinate until it falls off my calendar.

<shameless plug> Try http://www.instanote.io

I built it to scratch an itch especially for reminders. You can snooze reminders, for an hour, a day, or a week. I snooze them all the time until I’m ready to do it.

Oh, this doesn't have a web interface? Would be epic if it did!

I personally would open up my calendar, flip six months later, and then pick a convenient time, since there is wiggle room on getting something like that done. I'd probably add it for a weekend and add a notification to go off a few days prior.

In Google Calendar, it would less than a minute, based on a quick test I just did.

There are so many B2B apps that will do the first one. It’s called job management, and it’s a massive product category. But none of it is as simple as ‘download this app and take a photo’ because every business is slightly different.

And while your app could start so simple, and perfectly aligned with their use case, next they’ll want something else, and you’ll want more customers, so the customisation options begin. Until it’s so complicated to get it to do anything that it requires a consultant to set it up.

But will they really pay for it? Because at present they’ve chosen to tweet rather than google. And google + money would have solved this already.

> Lots of interesting ideas on here! Very cool site.

> Here's a pair of related tweets I found:

> > dear coders: I need an app that will scan receipts from contractors, automatically pull the relevant info re: business name, type of service, date, receipt #, etc, and file it in my phone contacts with a 1 year reminder on my calendar for future maintenance calls.

> https://twitter.com/LDBurnett/status/1379785147294310405

> > I need an app that’ll remind me when my cards are about to expire. Cards, passport, even products. Need.

> https://twitter.com/newii/status/1379439739615514627

> Together these might form an interesting app idea.

> Lots of interesting ideas on here! Very cool site.

> Here's a pair of related tweets I found:

> https://twitter.com/LDBurnett/status/1379785147294310405

> > I need an app that’ll remind me when my cards are about to expire. Cards, passport, even products. Need.

> https://twitter.com/newii/status/1379439739615514627

> Together these might form an interesting app idea.

That reminds me of https://grid.is/blog/your-biggest-competitor-is-a-spreadshee....

> I need an app that’ll remind me when my cards are about to expire. Cards, passport, even products. Need.

A tiny self mailing Google App script set to run daily and hooked up to a spreadsheet with expiration dates can do the job, no need for an app. Email reports are sent in case of any runtime crashes.

I spent a few hours building this. I’d love to hear your thoughts.

I think this is pretty cool and a great idea. One thing that I ran into within a couple minutes is that there are several entries that don't have enough context. It will just say something like "is there an app?".

This is awesome! It would also be really nice if the scraper would organize the ideas based on keywords in a given tweet.

I love this and have bookmarked it for daily review.

If I can make one recommendation, add a search function. That way users can find all the recommendations that touch on some field that's important to them.

For example, I'm working on an app for ebook authoring. It would be awesome to search through these for any ebook related app requests. It could help me to spot patterns related to my field, and give me ideas not just for apps but for features.

If I can make a second recommendation, make those search results subscribe-able... like a Google Alerts for Twitter-based app requests.

As someone who built a website called (roughly) "$5 business ideas" about 12 years ago this gives me great joy! It was mostly satire but still fun to build and populate with terrible ideas.

What's the tech stack behind it? Just curious, it takes me days/weeks to get any ideas out due to tech choice paralysis

Looks like Nuxt/Vue sitting on S3 with CloudFront as CDN. Presumably with some script on the backend to update the list of Tweets on S3.

/vaguely informed guessing

Yup, It's one simple lambda function that reads the JSON file from S3, searches twitter for new tweets since the latest, appends new ones and writes back to S3. Cron'd every minute.

This is awesome! What is the strategy to identify the idea tweets? It looks like some string matching, but I'm not sure.

Yes, nothing too special. There are 6 or 7 string matches e.g. "I need an app that" (minus retweets). But also retrieving referenced tweets.

You should go one level of abstraction further and make a site where users can aggregate arbitrary twitter search queries like this. AFAIK you can't actually "follow" a search query or hashtag on twitter.

A few hours? That is a fantastic return on investment!

By the way, you should definitely add this to your CV. It's the sort of technical thing that really stands out from other candidates.

Ah, they are catching on! https://twitter.com/dr0dil/status/1380230776017534980

"I’ve got to do a letter drop across 30+ local addresses. Is there an app that can plan the optimal route?" well,we still need a bit of research for this...

Getting XKCD/1425 vibes off this one :)

Same! :)

30+(a number smaller than, say, a hundred or two) is totally brute forcable.

I have a python script that plans walking routes between attractions at a large-ish festival that pre-calculates shortest paths by brute forcing dijkstra's algorithm between every pair of ~200 "intersections" in a minute or two on my laptop. (The app can then just look up the route by start/end location and draw it on a map)

Travelling salesman is hard, but even O(n!) is trivial on machines today where n is small. And there are simplifications and approximations that get something like within 1% of the theoretical shortest path in something like O(n log(n)) time (with significant startup/preprocessing time, but where N is between a few thousand and a few million these work).

Really? This is not a solved problem?

It's a very well known issue and it has been studied for a long time.

https://en.wikipedia.org/wiki/Travelling_salesman_problem

It's a classic NP-Hard problem. And depending on the size of the sample we can have pretty much the best route available, but we still don't have a way to solve this 100% accurately (particularly for large sample size)

> I need an app that's like @Shazam but for movies. Point your camera at the TV, press and hold, and it tells you the movie/show

https://twitter.com/call_me_trisha/status/138036131172963942...

This one seems like a fun project

I think you could probably accomplish this with just speech audio. After identifying the actors present, it would be a short jump to a list of possible productions, and given some dialog, you could further narrow it down.

Where could one find a database of actors with voice samples ?

It seems like most people need an app which tells them what apps already exist for the problem they are having. :D

https://en.wikipedia.org/wiki/Quixey

NLP has come quite a way since then, though the overlap with Google's functionality makes it seem like a losing idea regardless.

The one about saving your favourite shopping items from sites ... is called “add to favourites” and has been in browsers since 1990-something

Wow. That was a trip. People want some terrible apps. Why are so many people obsessed with how many people follow them on Twitter? Or how many people follow the people who follow them? Or who is unfollowing them?

A couple of these had me thinking "such-and-such can already do that... oh wait, Other Company bought them and we don't have that anymore..."

Yeah I think there’s also the case that people say they want and would use all kind of things but if you actually give it to them they realise they don’t want/need it or it has some flaw.

Nice app.

I also like Request For Startup (https://requestforstartup.co/)

Someone could make a good video chanel just live building shitty versions of apps from this list.

And someone else could make a competing youtube channel showing how 90% of them could be done just using a browser/bookmark, and 90% of the remaining ones could be done with stock phone apps like Shortcuts.

(And the only last one would be something like "I want an app that'll let me tether my phone so I can use data over my neighbour's Covid vaccine 5G connection.")

Imho there should be voting button (up/down) to see which app idea is the need of the hour.

Aha amazing! I've been wanting this for ages. I wrote about this exact idea on reddit a year or so ago but never launched

So props on launching!!

(mid way through this comment, ctrl+f for twitter for easy find) https://www.reddit.com/r/androiddev/comments/bvlv3q/does_any...

I paid for the 30x500 course (so far not worth it :-( ... it anyway the point of the whole thing is to first identify your audience, then find where they hang out online and then do this kind of research.

This coukd be linked up with Rand Fishkin's new thing where he has found a million people's twitter bios and linkedin bios and can therefore find "people who post a lot about sailing and work in finance".

But honestly, it casts a wide net and I think I need to niche down.

Why have you found 30x500 not worth it so far?

It was just a dump run of videos (I think they changed it that year). And it's me to be fair. But for the price I kinda expected more human involvement. I am taking another run with some more motivation now.

This is fun and super useful. Some low quality tweets could easily be discarded, though.

Hey, self plugging my project - https://needgap.com A problem validation platform which serves similar purpose as OP but bit more nuanced as the problem statements need to be well defined and the QC is high.

"Is there an app that can find the biggest dick near me that can fuck me good "

Sounds like someone wants an upgrade to Tinder/Grindr.

It's interesting to consider what the impact would be on society if Tinder had public reviews...

Or maybe Grindr. Do we know?

How would you sort that? Some combination of length, girth, rigidity, recent reviews, proximity, and a checkbox for circ'd or uncirc'd... the query logic is, shall we say, quite hard.

"Crawler" implies it follows links, this looks like it just performs keyword searches right?

Pretty cool and fun to read through.

If I can make one suggestion: make clicking any part of the Tweet card element go to the Tweet in context. That's the main action I will want to take, so it's a little annoying that it's relegated to one small link in the top right.

I don't think this happens frequently enough to make it viable... but

I think the other missing piece is App creators could register for problems that they solve and then your bot replies and says "There is an app for X <link to app store>"

In other words, you wish there was an app for telling you there is an app for something you wish there was an app for?

Is there an app for that?

Any social network, for finding friends.

"Ask people what they want and they will say a faster horse" - Henry Ford

Enriching the responses with demographic data, extracting interesting features with NLP and then clustering against those features could give you some good insights.

I.e. Millennials in Belgium are not happy with their toy stores.

Ok panopticon

There's a problem, it's down. I mean I'm getting a network error.

Can we have RSS there, please?

"i need an app that will tell me how to find weird grouchy septuagenarians in my area who have no desire to be romanced so that i can seduce them when they least expect it"

Nice. It also catches things that are not app ideas, but interesting. Like:

"Is there an app that you hate but always use?"

YES! Instagram. I hate the app. But I have to use it for business purposes.

Wait long enough and you'll see some kid at a hackathon posting "I wish there was an app for splitting the bill with my friends..."

Cool idea! You could have a simple up/down arrow voting below each idea to get some interactivity.

NLP would make this much better, try to categorize and provide some statistics what people are talking about!

Ideas are the easy part.

But now you have at least one person with a non zero chance for betaing your app for free!

Awesome. I always tweet al the wishes and so should you.

I get a network error.

can someone make an app that’s actually fun to go on

Yeah, pretty much

Reply on news.ycombinator.com