marcusb

I had a customer once who would just absolutely berate people over email for the tiniest thing. Totally unbearable and unreasonable. So, whenever he would go off, I'd tell him 'look, I'll be in the area [this afternoon|later|whenever]. You going to be around if I stop by?' Any conversation with him that could be deflected to an in-person discussion could be peacefully resolved in short order. Trying to convince him of anything over phone or email was an exercise in frustration control.

I heard somebody say at a conference one time, talking about how much more productive in-person meetings are in reaching agreement, "there's a lot of bandwidth in a room". I think there's a lot of truth to that.

0 - ironically, this was at a ISP network engineering conference

I've never (to my knowledge) had a review on Amazon rejected, and I've left very some negative reviews, including when I received counterfeit items.

I always thought the review scams on Amazon were more driven by the third-party sellers doing stuff like listing takeover, astroturfing reviews, bribing customers for good reviews, etc., but maybe I'm wrong. I have personally received multiple offers from third-party sellers of incentives to leave good reviews.

> Ok, maybe very niche hobby products, but then I wouldn't expect a warranty.

I bought a bunch of parts for a racing drone from Aliexpress because I didn't expect a traditional retailer's warranty to really matter much. ("This frame has been in a crash. No warranty.") What's the point of paying extra in that scenario?

I haven't tested that behavior, sorry.

I've seen a bunch of requests with forged ChatGPT-related user agent headers (at least, I believe many are forged - I don't think OpenAI uses Chinese residential IPs or Tencent cloud for their data crawling activities.)

Some of the LLM bots will switch to user agent headers that match real browsers if blocked outright.

The only thing that would make that even more perfect would be if there was some way for the site owner to signal to prospective bots which parts of the site are open to the bots to visit. I know this seems really complicated, but I really think it could be expressed in a simple text file.

I once had a coworker who called this "bullshit bingo" and had a bingo grid drawn on a whiteboard at her desk with all of the latest buzzwords.

On a somewhat-related note, my grandfather told me that while he was in Officer Candidate School in the Army, there would be someone assigned to ring a bell whenever a person who was leading a briefing or otherwise presenting faltered with an "oral pause" (uh, ummm, etc.) I don't know if this was a normal or ongoing practice.

This was more than a decade ago, so I doubt they care that much, and if you want to accuse me of being a liar, please have the integrity to just say that.

First, I'll say it is pretty common to use the terms interchangeably. I don't think anyone was confused by what I wrote, or that your "clarification" was in any way helpful. See, for example, these legal groups using the term "protected group" in relation to US employment law:

* https://www.osbar.org/public/legalinfo/1095_DiscriminationEm...

* https://pedersenlaw.com/practice-areas/discrimination/

Secondly, I used the phrase "protected group" referring to disparate impact, and here, your assertion (to the extent it has any validity at all) is simply incorrect. The entire idea is to ferret out subtle acts of discrimination that have an outsize impact on a group consisting of members of a protected class, and in the case law you see the phrase "protected group" used explicitly. For example:

  On the contrary, the ultimate burden of proving that discrimination against a protected group has been caused by a specific employment practice remains with the plaintiff at all times (Watson v Fort Worth Bank & Trust, 487 US 977 - Supreme Court 1988[0])

0 - https://scholar.google.com/scholar_case?case=637945611431669...

All of the RSU offers I ever got stipulated the grant was "subject to the approval of the board", i.e., not guaranteed. That said, I'd be absolutely livid if something like this happened, and would be expecting my manager to either make it right, or I'd look for a new job at the first available opportunity.

You can't do good business with bad people.

People karma-farm here for no obvious commercial benefit. I'm sure the activity would be greatly reduced, but not eliminated, with a lack of commercial incentives.

I took it as a poor-quality local model. I suppose it is possible, but it seems...unlikely... that one of the three responses to that post was a person who was so upset by the idea of serving junk content to scraper bots that they registered for an account, wrote their criticism, and then disappeared from the site.

At that time, there were a whole bunch of flagged-dead comments from newly-created accounts that had compound-word usernames (such as the text I linked to above from 'earlydeveloper'.) So, if these are people manually writing those posts, I understand the behavior even less than the bot hypothesis.

I can understand what the bots posting this stuff on reddit are after, but what puzzles me a bit are the posters here who clearly are LLM-backed bots that post once or twice without any affiliate links or other visible scams, then disappear. Maybe they are getting banned but it isn't obvious? (If so, good job dang and team!)

For an example mixed with a bit of irony, a few months ago, I submitted a link to a content obfuscator (meant to target site scraping bots) that I wrote. One of the replies was from a brand new account, that hasn't posted before or since, with a fairly obvious LLM take:

From https://news.ycombinator.com/item?id=42517774:

  If your content got ingested by scrapers that don't respect the robots.txt, but that it was copied to another domain with a more lenient robots.txt, you could poison legitimate datasets.
  It seems wasteful to actively try to sabotage humankind's technological progress.

A web-based implementation of the Arca Musarithmica is available at https://www.arca1650.info/

> layoff plans must be communicated ahead of time. Minimum 30 days notice, usually much more

In the United States, employers with more than 100 full-time, non-probationary employees must provide 60 days notice of most planned layoffs[0]

> - LIFO principle for layoffs, newest employees are let go first. Stack ranking not possible

This is functionally equivalent to a stack ranking in that it is a forced-distribution scheme. It is just based on a single factor that is outside of the employee's control. Say what you want about stack ranking, but people do have a large degree of control over their job performance.

> Any kind of discrimination is forbidden

In the United States any kind of job discrimination against members of protected classes[1] in illegal. Even inadvertently disparately impacting[2] members of a protected group is illegal.

0 - https://www.dol.gov/general/topic/termination/plantclosings

1 - https://www.eeoc.gov/employers/small-business/3-who-protecte...

2 - https://www.law.cornell.edu/wex/disparate_impact

> Usually there is a hidden variable that you don't know.

This is frequently the case. I've worked at big employers (comparable in level of corporate-ness to Google if not absolute size) where the layoff process, roughly was:

1. Aggregate layoff target gets set and apportioned amongst functional leaders, then targets cascaded down to the line manager level.

2. Managers fill out a stack ranking spreadsheet for their team across a few metrics including a boolean "diversity" field[0]. There were many rumors about the "diversity field", most notably that anyone so flagged would not be fired, but so far as I could tell these were false (see point #4)

3. People to be fired are developed based on these lists (I.e., if a manager has to fire two people, then the two lowest-ranked employees per the spreadsheet are selected.)

4. HR does a meta-analysis of all to-be-fired employees, ensuring that a disproportionate number of employees from protected classes are not impacted. If too many are, then some of the next-lowest-ranked employees are selected to be fired in their stead.

As far as I could tell, the only part of the process where any sort of individual, human consideration was occurring was maybe at the line manager level if they decided to tweak the stack rankings based on who they felt deserved to be protected. And then, to the extent that happens, you have all the problems with bias and favoritism that come into play.

0 - I realize this is probably controversial, but I saw it with my own eyes.

You and me both.

To be fair to thunderbird120, the author of this piece made edits at some point. See https://archive.is/K4n9E. No discussion of the recent TPU releases, or TPUs for all, for that matter.

> SVG complexity is a good counterpoint to “why this instead of SVG”. But I don’t think we need to even address the claim that most tools for SVG generation use JS, if the concern is JS on the client.

If you want to address the generation of charts on the client, then, yes, JS is something that needs to be addressed, because SVG (or Canvas) are not in and of themselves systems for generating charts. There isn't a <PieChart> primitive within the SVG spec, to my knowledge.

The entire purpose of this library seems to be generating charts in pure markup without any scripting language dependencies. Trying to pretend that isn't relevant, or suggesting that "SVG" is, in and of itself, a viable substitute is just really weird.

Maybe they read your comment?

Yes, I understand that. But it does require some method of creating the SVG data, which is generally going to be more difficult than styling td elements with `--start: 0.2; --end: 0.4;`. Most extant libraries for this use JS.

How would you do that client side in pure CSS? "No JS" seems to be a goal of the project.

Isn't hoarding data for training purposes a key part of OpenAI's business model? I get that they don't have a reputation for selling that data (or access to it) yet, but, what happens if/when funding dries up?

I definitely don't trust Google -- fool me once, and all -- but to the extent I'm going to "trust" any business with my data, I'd like to see a proven business model that isn't based on monetizing my information and is likely to continue to work (e.g., Apple). OpenAI doesn't have that.

From the article:

> I’m forgetting something. Oh, of course, Google is also a hardware company. With its left arm, Google is fighting Nvidia in the AI chip market (both to eliminate its former GPU dependence and to eventually sell its chips to other companies). How well are they doing? They just announced the 7th version of their TPU, Ironwood. The specifications are impressive. It’s a chip made for the AI era of inference, just like Nvidia Blackwell

What is a distribution but a collection of software packaged in a particular way?

WNT = V + 1, M + 1, S + 1

Z Xplore at least used to include Z series access.

https://ibmzxplore.influitive.com/users/sign_in

Related:

* Framework temporarily pausing some laptop sales in the US due to tariffs(https://fosstodon.org/@frameworkcomputer/114297967333461078) 170 points|leotravis10|15 hours ago|34 comments

* Framework "temporarily pausing" some laptop sales because of new tariffs(https://arstechnica.com/gadgets/2025/04/framework-temporaril...) 19 points|pbui|16 hours ago|4 comments

* Framework pauses US sales of some laptops due to tariffs(https://bsky.app/profile/frame.work/post/3lmajsf6uws2p) 56 points|m-hodges|17 hours ago|0 comments

* Framework pausing sales of certain laptop models due to tariffs(https://twitter.com/FrameworkPuter/status/190930422507239438...) 11 points|jljljl|17 hours ago|0 comments

That's surprising to me. (I won't be like the other commenter and say it is "shocking" or "bizarre". Just surprising.)

To answer your original question: I worked exclusively with Fortune 500 customers, as a vendor in the security and network infrastructure spaces. The customers weren't in any particular vertical, but included healthcare, utility, and finserv shops (I.e., risk averse businesses.)

I really didn't expect my original comment to be that controversial. In fact, it is almost tautological.

If you've worked successfully in sales, you have to understand the concept of a compelling event - if not by that specific name. In BANT, it's split between need and timing. In MEDDIC, it's "metrics" and "identify pain". Whatever you call it is immaterial, the substance is, effectively, in order to get the deal done, what change does the customer need/want, how important is this change to them, and what happens if they do nothing (put another way, can they avoid change?)

In product management, it is similarly a pretty basic element of market analysis to identify what features would be compelling to current and potential customers. You might bristle at that characterization of PM, so feel free to make it sound more strategic and thought-leadery, but that's basically the job.. Did you ever manage a product that was in a net new category, but where you knew you would be competing for and need to capture existing budget? Did you try to identify what compelling benefit ("change") you could offer potential customers to capture that budget? I sure hope so.

If you still don't like that, and are still in PM (or sales!), then have the courage of your convictions and on your next customer call to latch on to a Digital Transformation project, try focusing your pitch on 0% change and 100% stability. Let me know how that works out for you.

Underlying all of this is a simple truth: people buy to get some perceived benefit they don't already have. That is change, by definition.

If I were a clothing salesman, and you a clothing purchaser, and I tried to sell you the same outfit you were wearing when you walked into my store, you'd think I was nuts.

Now, reliability is certainly a thing people say they want, and something they weigh in their purchasing decisions. You'll get no argument from me on that. I sometimes acquired customers due to reliability problems they had with competitors, but that was extremely rare. It was something customers threatened more than they actually did. Devil you know, and all. Ironically, this is one situation where the fear of change usually prevails, at least in my experience.

What did happen -- regularly -- is I had customers who ran a full RFP process upon lifecycle refresh. Even if they had a solution that worked great, they were expected to survey the product landscape and make sure they were still buying the "right" thing, for some definition of "right". To the extent they swapped products on refresh -- and they sometimes did -- I don't see how one could describe that as anything but a bias for change over stability.

Another thing that happened regularly was analyst-induced/customer-championed Resume Driven Development. Healthcare company that runs everything on an AS/400? Better not get left behind and move to the cloud/microservices/fad of the week! Change > stability.

But, here's the thing: any change in condition, whether user visible or not, potentially impacts reliability. Simple expansions fail, due to scaling thresholds and other factors. Version upgrades fail - even narrowly focused patches introduce unexpected regressions. Net new and tech replacement projects in IT have abysmal success rates[0]. (And all are change.)

Nobody goes into these projects expecting failure, but that is very frequently what they get. Every single one of my customers, when asked (and we did ask!) would place stability as their #1 or #2 goal, yet they were constantly replacing stuff. Sometimes for good reason, sometimes for reasons I could not understand, sometimes for reasons I disagreed with, occasionally to address a legitimate reliability issue.

Just like the old investing adage that "price is what you pay, value is what you get," in enterprise tech, you can change the punchline to "change is what you get". It is what people are buying, whether they realize it or not.

0 - various studies: https://calleam.com/WTPF/?page_id=1445