michaeloblak

All SaaS survey products are insanely expensive. It’s always $0.04-$0.15 per response.

SurveyMonkey - $0.035, TypeForm - $0.045, HotJar - $0.08, Survicate - $0.179

It’s saving *a single record in the database* after a request to the API. Plus wizard for creating the form. Then download the CSV with the results. Also, all companies are charging on a monthly basis. Most surveys goes from time to time. I’m talking from my experience, but there are months where I need to survey 100 users and months where I need to survey 30k users. Why it’s not billed on a ‘per response’ basis?

Do most companies write their own survey software, is there anything cheaper or are you using some open source?

Is it possible? Let's say Alice has 2 GPUs idle at the moment, Ben has 1 GPU, and Chris needs 3 GPUs for the next 12 hours. How to build such a system, and what problems there might occur? How to handle turning off one's machine? Does it even make sense to run training or inference on such (latency)?

What software do you use? How much time did it take your team to set up a CD? How much time do you spend on maintenance? How big was your team when you set up a proper CD?

Daily I'm working with a lot of SQL, mostly in a self-hosted Redash. But, I caught myself on commenting some part of a query and then writing another query below the commented-out code. I thought it would be awesome to work with data in a Jupyter-like fashion - then I could tell the whole story, why I went this way with research, why such a query. So I added SQL support to my Jupyter instance, a started working with data in Jupyter.

Now I can tell a better story and I have the full history of "why I did this". But it feels somehow not cool, also keeping the credentials to the database inside the notebook - not cool. _(I know I can keep them in environment, but that's not the point)_

Is anyone querying SQL from Jupyter? Is there a better way to query data, and query the results with another SQL? Then use Python to do some stuff on data. Then plot the results. Keeping all the mid-steps at the same time.

Is there something like this? Would that help with stale PRs?

Why there's no voting on Arxiv? Or even the number of the page open or PDF downloads? It may help filter out really good papers. Citations count is a rather slow indicator, because well, not many people cite you after a month or two. I'm asking because I want to be up to date with all the significant research from my field (I'm an engineer, not a researcher), but there are 50+ papers published every day in my area, which is incredible. But not all of them are worth reading.

HN like voting would do.

Let's suppose there's a company is closing medium size enterprise deals. This company knows exactly what it's doing, how to sell, who they need to target, how to target them etc. This company has more leads than the capacity to handle them by the in-field salespeople.

I saw that there's a big aversion from the sales side on to setting up more than one person with the same area. They are afraid that the other one will take good leads. They are scared that they will close all the leads, and lose their jobs.

From the company point of view, it's making no sense. It's obvious that it's better to close all the deals fast. Or is there something that I don't see?

How good/big/fast companies are handling this issue?

Today I have such situation:

1. Listening to Spotify on the desktop in the app

2. Suddenly my music stops, and I see it's streamed on some different device - unknown for me.

A guy sitting next to me in the office has the same thing with his Spotify account. We are both family accounts owners. We both change passwords and logged out all devices. It didn't occur again yet.

Does anyone have a similar situation?

Btw. the worst thing in this whole situation is that person was listening to really not my kind of music :-)

I'm just curious if there's any Open Source JS handling all events and the back end which can handle them and provide some UI rather than querying the DB of events itself.