robflaherty

I linked this post because it explains how the lottery works which I thought was more interesting and relevant for HN.

Always liked Reinertsen’s post on reframing tech debt to deferred work.

“When we refer to postponed work as technical debt this automatically biases us to assume that both ongoing and future costs are more certain than they really are. This, in turn, causes us to overestimate these costs, leading us to be overly cautious about deferring work. If it is your intention to bias the decision against postponement, this is clearly the best term to use. However, if you are trying to carefully weigh of pros and cons of postponing work, I’d recommend using a more neutral term like deferred work.”

http://reinertsenassociates.com/technical-debt-adding-math-m...

I see this with Framer sites all the time. Happens in Chrome too.

I know the intent is good (it is!) but this common notion that good design is a toolbox of fun little games really prevents the design function from getting taken more seriously. In the business world at least it ensures our seat at the kid’s table.

https://archive.ph/vb6IB

Is there a reason we won’t have LLMs that only speak in JSON? These JSON hacks are clever and cool but feel like they’ll be obsolete in 6 weeks.

The OpenAI Cookbook has had an easy to use chat with a PDF app example for a few months.

https://github.com/openai/openai-cookbook/tree/main/apps/fil...

Worth remembering that priority and sequence are not quite the same thing.

There's a useful framework called RICE where you evaluate/score ideas by Reach, Impact, Confidence, and Effort. Sometimes assigning numeric values feels forced so I often use it as "dimensions to be thinking about" rather than a literal scoring system. [1]

There's some great stuff in Reinertsen's Flow book about different sequencing strategies and when to use each. Eg, When delay costs are homogeneous, do the shortest task first. When task durations are homogeneous, do the highest cost-of-delay task first.

[1] https://www.intercom.com/blog/rice-simple-prioritization-for...

The author's framework-"you just start small and only add one at a time as needed"-and given example illustrate exactly the process that leads to too many metrics. Pick a few metrics, add another, add another, add another. If you have 5 teams working on a product you quickly end up in the 50 metrics scenario.

Too many metrics can be a problem but it's not the real problem. The real problem is choosing metrics without any regard for the decisions they're supposed to inform.

When you understand that the purpose of all measurement in business is to reduce uncertainty for a decision to be made, everything comes into focus, and you'll have a natural constraint on your scope of measurement.

According to the ePrivacy Directive, they should not last longer than 12 months

The quote you keep referencing is false. The ePD says nothing about this.

The ePD text says nothing about a 12 month cookie expiration and also ePD != GDPR

It doesn’t. Or if it does, no one here has yet to explain why.

Can you explain what is invalid about it? The list of functional cookie permissions on the consent opt-in specifically mentions cookies to remember log in details.

“Zoom cookies are firstly written when the user connects to the website zoom.us and accepts the cookies options.”

That was the moment Zoom received your consent to store data transmitted by cookies. Adding a few more cookies to the pile, regardless of expiration date, doesn’t change the agreement.

Rummaging round the cookie bin on uninstall is a nice find and deserves a raised eyebrow but this doesn’t really have anything to do with GDPR.

Based on the replies to the tweet very few have assumed this is the case and no, “using as a CDN” and “accidentally linked to” are not the same thing.

"I'd assume that for banks all code that goes into production gets audited"

The point of tag managers is to let non-developers update production without going through a code deployment so the usual code review and oversight doesn't apply.

Also worth keeping in mind that this is the bank's marketing site. It's almost surely managed by a different team than the actual online banking site and probably has looser restrictions.

The Internet Archive rewrites contents of scripts to inject the archive URLs. A better explanation than OP's clickbait is that someone went to the archive to copy/paste misplaced tracking code.

GDPR is not a cookie law. You don’t need consent to run out-of-the-box GA with IP anonymization.

Today’s internet is far more creative than the internet of the Flash era. Creativity has moved on to more consequential things than graphic design.

For iOS Refine blocks Taboola/Outbrain https://apps.apple.com/us/app/refine-customizable-ad-blocker...

I think only one person said it was cool :) What is disturbing about it? Do you feel it's a privacy violation?

I haven't tested it with infinite scroll but essentially it'll be a race. If the user can scroll to the bottom of the page before infinite scroll is able to load new content, the 100% event will fire. Otherwise it'll just register the previous scroll event (75% or a specified DOM element).

All engagement metrics are pretty loose. Scroll depth percentage doesn't tell the whole story but to the degree that scroll tracking is useful, I think percentage provides a simple, document-independent metric.

Btw, tracking scroll points for specific DOM elements is one of the plugin options, so for instance you could track the end of the .comments div as you mentioned.

Hi Brad,

Just wrote you back! You know, I actually don't see major publishers embracing responsive design, at least in the short term. I think mobile-specific sites and dedicated apps are much more inline with how they see the world. It's also a business opportunity for them to sell separate experiences as separate products, right? Merging everything into one isn't necessarily a win.

Hi, I'm the author of the post. The examples you mention are server side solutions requiring some form of server side detection. You're talking about serving ads to separate mobile and desktop sites.

The goal with responsive design is to serve the same site—the same mark-up, the same assets—to all devices. Which means that any conditional serving of ads has to happen client-side.

Sorry about that! It's fixed now. That was a mobile-specific style that I'd accidentally applied globally.

I agree that a comprehensive test would include the optimized scenarios that you mentioned. In this case the intent was to show that even when comparing two extremes - a very unoptimized page and a super-optimized page - the results were not as disparate as we'd expect. Scenarios involving sprites and domain sharding we'd probably expect to fall somewhere in the middle. Thanks for your comments.

Hi Jeremy, I'm the author of the post. The tests were not conducted locally, but over a real network, albeit a low-latency one. Latency is definitely a key factor, so I'm planning to run some more tests under different network conditions. Thanks for your feedback.