Hacker News Reader

Lots of "Live chat" customer-support type systems do it this way. You can see what your prospective customers are typing as they type. Every character, backspace, etc.

A bit creepy and violates privacy expectations, I think.

What. Someone should do something about this. I absolutely am shocked by this.

It is really terrible. https://lifehacker.com/customer-service-chat-services-see-ev...

Do what? I would have been surprised if this wasn't the case. Websites see every keystroke you make while their targeted. People should realize that. If they don't, they're in for some rude surprises.

The browsers could buffer the text in text fields until the field loses focus (clicked another field or submit) or hit enter. It would break search-as-you-type functionality, but not the biggest loss in the world.

devs would just catch keyboard events and simulate text fields you're not getting this one back in pandora's box

There's always the option of typing your stuff in another program entirely and copy/pasting. Sure, not everyone would be willing to do that, but it's really not that inconvenient if you know your keyboard shortcuts (ctrl+a, ctrl+c, alt+tab, ctrl+v in that order, easy for a competent computer user). And socially anxious people may be more comfortable with that overall.

Copy+paste won't trigger keyboard events to my knowledge. So if the page was just simulating a text input field, paste wouldn't work. (Which would cause issues e.g. pasting account numbers to be fair.)

There are quite a few browser extensions that make this easier. The best is Pentadactyl, a complete UI overhaul for vi bindings everywhere that lets you hit Control + i to open any text field in any (blocking) external editor (gvim -f by default), so you can type out your message and then type ZZ to send all the text to the field at once. However, it’s only for XUL browsers. I think there are also WebExtensions that offer similar external editor support, but when I use browsers limited to them I use wasavi, which pops up a javascript-powered vi clone inside the browser (but still only sends the text down to the area when you save). IIRC SurfingKeys also has a similar thing, with a codemirror-based editor that’s IME more limited than wasavi. Of course none of these work on exceptionally complicated things like Google Docs, but they often work on simpler WYSIWYG rich text areas, and wasavi and maybe SurfingKeys even support converting the HTML within them to Markdown while editing, and back when you save.

How about disabling all network actions until user hits send button. I know very complicated to implement.

receiving raw keyboard events should require special permissisions then like audio and video.

Maybe we can achieve that when we create the next internet.

Browsers already have good UX reasons to do something like this. They should offer version history on textareas, for example, so you don't need to ever worry about losing state.

But they don't. So I build it (draft system) for my users by saving what they type as they type it, and they are very thankful.

If you would kill my ability to build that feature, then you better replace it with a solution.

For a site being creepy that way it's usually shoved into a database somewhere. Not shown to a human.

Yeah, tools like Salesforce/etc. make this absolutely trivial.

I once realized this was happening when the customer service rep was replying to me almost immediately after I hit send. Initially I was like how the hell did you reply/type this fast, but it dawned on me after a few more messages.

I used Verizon's recently, and it was actually forthcoming about this. I think it was part of the text content shown when the connection began (after the annoying pre-chat Q&A during which they try to convince you they can solve your problem without involving a human).

Not sure I get the outrage to this one. Do you anticipate typing private info, backspacing, then replacing it with some bland query about switching internet plans? The only time I might have minded true real-time chat was in grade school talking to girls I fancied, but other than that, it makes a lot of sense. Saves time for the support agent and if they get a computer illiterate customer typing at 1 WPM who doesn't know to use enter to send, they can at least try to help.

Now, if it were a normal web form, I'd feel differently, but if I've opened a support chat session, it's because I want to send the things I type to support.

> Do you anticipate typing private info, backspacing, then replacing it with some bland query about switching internet plans?

I anticipate people accidentally pasting something from their clipboard that could be private/sensitive.

Or typing an irate and cursing message that they don't intend to send, deleting it, and then sending a polite message based on the first.

This might work in the customer's favor though as the rep might offer concessions to try and get back on good terms

Or they get irate at how dare they secretly be irate at me, I'm sick of all the irate people!

Or they may decide to put in minimum effort or just close the conversation saying "I'm glad we were able to help you."

I loved this feature when I first experienced it in Google Wave. I don't think it belongs everywhere, but exactly as the article author says, it makes the conversation feel smoother and more like I am actually talking to someone in real life because I can think about what they are saying as they are typing it.

In a real life conversation, you digest in real time as you hear the other person. You don't have to wait for chunks of their thoughts.

(Talking about Google Wave) I liked it in theory! In practice it felt super awkward and almost privacy-invasive, but I think that's because we weren't use to it. If it became commonplace, it would ultimately seem more natural.

All it did was have two people ditch what they were writing and perpetually respond to what they thought the other person was writing.

When it worked out well though, there was an incredible feeling of being in sync with the other person.

I never used Wave much, but I'm old enough to have used (the shell program) Talk many, many times to communicate with my friends on the university system. I really hated the live typing feature. In order not to seem rude, you had to focus on the conversation so you were ready to respond as soon as the other person was done. You had to be patient with the typos and slow typing of some people. And even if the deleted parts of the conversation had relevant information, you had to at least pretend to ignore whatever was typed & deleted. It had all the limitations of online communication (no facial cues, no body language, no tone of voice) but didn't have the good grace to be asynchronous which would let you do more than one thing at a time. It was so frustrating to be stuck in one place staring at letter after letter slowly appearing.

It felt like all the disadvantages of an in-person conversation without any of the advantages. It was such a relief when I finally convinced friends to rely more on email or chatting in places like MUDs.

And sorry to rant at you a little. It's just I can still remember the frustration vividly after 30 years.

No need to apologize. That's a totally valid and understandable take on it.

I loved this feature in when chatting on bulletin boards. So much so that I wanted to recreate the feeling: http://sysop.chat/

The live feed of talk/ytalk/etc was a mixed bag. If the person you were chatting with was a decent typist it was great as you could process what they were writing as they were typing.

But much more often the person on the other end was super slow, constantly making and then slowly fixing their typos, etc. I spent most of my time getting frustrated watching them muddle through

This is how it used to work on BBS's.

I had a 1200baud modem connected to my C64, and some of the time I could type faster then the transmission speed of the connection. It would be 'fun' to be able to type so fast that the computer would end up buffering the letters..words...sentences and I could watch what I had typed scroll out to the terminal. This was the single biggest motivator in learning how to touch-type.

I do know that sometimes the connection was just so poor that it wasn't because I was awesome, it was just a shitty connection. But I never let that truth get in the way of my ego. =)

I wonder what kind of overhead you were dealing with there. As raw characters that would be at least 120 per second. If you send a TCP packet every second it would hold 80.

Maybe the connection was bad sometimes but would it really have been so bad that an already-low-frequency modem consistently ran at less than 10% speed? My mind races thinking about the ways to configure a data stream to lose so much speed.

> My mind races thinking about the ways to configure a data stream to lose so much speed.

Not quite the same peripheral, but the 1541 disk drive was designed to be backwards compatible with older Commodore models. As a result, it had a very low read and write speed compared to other computer models. I wonder if it was a similar case for the modem.

> in certain situations, a modern version of this would actually be pretty nice

I’ve done something similar before in Google Docs, and recently in CodiMD/HedgeDoc (which is much nicer and simpler, and doesn’t have comments except for html ones, reveal.js slide notes, and opt-in Disqus on the published doc). I think Etherpad could be used almost identically to talk, except with color coding differentiating each user instead of panes.

Email is all of that disabled.

(as long as you your client doesn’t load tracking pixels wrong) Edit: the article: “Gmail tells you if your friend is typing, and whether there’s any text left in the box.”

Doesn't Gmail load and proxy all external images anyway? Am I remembering wrong?

Of course, not everyone uses Gmail, but if it loads all images whether or not they're opened, tracking pixels are much less useful.

IIRC their servers cache them the first time you open an email, only preventing them from triggering again if you open the email more than once

EDIT: it looks like it doesn’t even do that, and “might even benefit email marketing, for sure not blow it up” if this isn’t out of date: https://blog.filippo.io/how-the-new-gmail-image-proxy-works-...

I wish email supported this at the corporate level.

It wastes time when someone sends an email to a team of 5 people and they’re all independently working on the same reply.

If you want to hide the writing process, there's always email (and other solutions).

I will always ALWAYS prefer the instantly-visible option. In a world devoid of vital visual contexts like vocal inflection and body language, a visible "stream of consciousness" is so meaningful.

There's a steadily-growing norm where people freely ignore instant-messages if responding quickly doesn't benefit them, but sometimes the message sender has no other communication recourse for an important and time-sensitive topic, and if you have no idea if they have even read your message then it's very frustrating.

There are forums that support this functionality and it's actually kinda nice (but hectic with too many people in a thread).

Which ones for example?