Tournament username: Touchgram_Andy
Karma: 235
Recent posts
Shared by AndyDent-Touchgram · 113d ago · 5 comments
Shared by AndyDent-Touchgram · 118d ago · 4 comments
Shared by AndyDent-Touchgram · 140d ago · 14 comments
Recent comments
AndyDent-Touchgram · 2d ago

I think it's achievable given modern ML but still probably harder than most people think. is providing suggestions based on comments and code. Friends who have tried it say it delivers wildly varying quality.

So I think we will nudge up to this with increasingly smart assistants but the problem with that is it still needs someone in charge, who understands what's happening well enough to know the assistant screwed up.

That someone in charge may be another ML model in future.

I've thought about this stuff, off and on, a lot - there are things which most codegen/design tools still miss completely. I just don't have a clone of me to pursue those ideas.

AndyDent-Touchgram · 3d ago

Descriptions in English has been a holy grail since COBOL was invented.
By itself, it's never gonna work, because it's a very sloppy way to specify things.
I think it highly plausible that an interview-based process which proceeds from rough description and has a bot refine its understanding, will work for most cases.

This would mix well with automated codegen from rough sketches such as AirBnB are trialling

I see this as a variation on the lesson that Palm learned about recognising handwriting on very low-power systems. At the time when the Newton was being mocked in national cartoons, they had the insight that a meeting halfway would work.

So Palm introduced a simplified writing system which was:
1. easy to recognise
2. easy to learn to write in with a stylus.

AndyDent-Touchgram · 5d ago

I can understand why it's limited to text right now but that alone is going to prevent me doing more with it. I'm already using Evernote so could just use that as a backend via the API if that was all I wanted.

AndyDent-Touchgram · 5d ago

I'm more of a real programmer than most of the people on Pioneer and I don't dismiss these environments.

AndyDent-Touchgram · 6d ago

Thanks for the heads-up but I'm looking for a web app solution, not a native app. I don't want this to be on an app store because if I monetise it, it will be at a low level. Plus it's more suited to be a web experience - some uses will be more convenient via laptops.

AndyDent-Touchgram · 6d ago

Thanks, Ken

My priority is being able to iterate on a great UX with scalability as a lower priority.

This is not a major project for me - it's a side-project for something I need personally. It has some potential to turn into a much bigger app but if there's a big barrier to entry it just won't happen.

I'm also old enough to have a very cynical but also realistic take on dev tools and low-code - I go back to before dBase II in the 80's,

AndyDent-Touchgram · 7d ago

Thanks, that's useful to know but their pricing is probably too high to build my own service on top. If people were already using and paying for Notion it may be viable ($4/person/month). So far, as you'd expect, the API is designed to help you "build interactive experiences for users within Notion."
API doc says "Only text-like blocks are currently available" so that rules it out for now.

Yeah, I'll keep this in the back of my mind - it might just do the job once expanded and just selling to Notion users is not an unreasonable size market.
From first glance, the API is more like the one I've dabbled with years ago in Evernote - it lets you access the data within the service. I'm not yet seeing anything that lets you build an interactive experience running within Notion itself, as a different kind of block.
quick search found explainer video - interesting that they build an "app" for it in Glitch!!/block-x-block-final?

AndyDent-Touchgram · 8d ago

Disagree somewhat - I definitely do store my attention for later - I clip and bookmark a lot of stuff from an extremely fast scan, then review when I'm watching TV with my wife.

That implies some things about timeliness.
1. My attention is gained for seconds initially but I'm not engaged in realtime.
2. The material may get significant attention but not for days/weeks. It also will possibly continue to be forwarded at odd intervals, that's why I clip much stuff.

AndyDent-Touchgram · 13d ago

I noticed a few of the very generic comments a while back and emailed Sam about it. Some of them were very obvious - I save all the feedback I give and receive so I could quickly go back and check them. In particular, I noticed them giving identical feedback on other's status reviews, when I went in to write followup comments to responses.

AndyDent-Touchgram · 15d ago

Pretty sure no authority would be bothered by the Pioneer cash - it was based on determination of merit with very little element of chance involved (you could maybe argue the element of chance in that some of your rise on the leaderboard is determined by who you get to review).