marcusb

This reminds me of the story a few days ago about "what is your best prompt to stump LLMs", and many of the second level replies were links to current chat transcripts where the LLM handled the prompt without issue.

I think there are a couple of problems at play: 1) people who don't want the tools to have value, for various reasons, and have therefore decided the tools don't have value; 2) people who tried the tools six months or a year ago and had a bad experience and gave up; and 3) people who haven't figured out how to make good use of the tools to improve their productivity (this one seems to be heavily impacted by various grifters who overstate what the coding assistants can do, and people underestimating the effort they have to put in to get good at getting good output from the models.)

Anything is possible, but he didn't start the conversation about cheating. Someone else brought up something to the effect of they thought game shows were fake, then he told his story and a third person the table searched for and showed the video.

> Apparently he gets pretty high with friends and shit talks - but when when the search started to suggest some pretty way out things along the same lines, he landed that their conversations weren't private any more.

I had an experience like this several years ago. I was having dinner with a customer, and one of the guys brought up this story about how he went to school with someone who got caught cheating on Who Wants to be a Millionaire. Later, back at my hotel, I pulled up YouTube and the first recommended video was of the guy who got caught cheating on the game show. I had not searched for this during the conversation (or prior) nor do I watch game show videos on YouTube, or cheating scandal videos on YouTube.

Here's what I think happened: somebody at the dinner googled it, and the video got recommended based either on geo-location data (we were in close proximity) or because the person who googled it was in my phone contacts, or maybe both. But, I don't think Google/Youtube was recording anyone's conversation to make that recommendation.

It's still around

I think you're bending over backwards here to rationalize this 'review'. Assuming this person does use throwaway accounts as you suggest, you'd have to believe that in February of last year, they were busy writing a dice-guessing game and a rock-paper-scissors implementation and in February of this year, they are experienced enough to write a lengthy critique of the architecture of someone else's project. A review, by the way, that is longer than the code it criticizes.

If you believe that, that's fine. I don't.

As for the correctness of the 'review', it is absolutely nuts. Total nonsense.

The current models really seem to struggle with the runes...

The original code is explicit as to the intended purpose:

        # Auto generate random X-Forwarded-For if doesn't exist.
        # Otherwise AWS forwards true IP address in X-Forwarded-For header
        x_forwarded_for = request.headers.get("X-Forwarded-For")
        if x_forwarded_for is None:
            x_forwarded_for = ipaddress.IPv4Address._string_from_ip_int(randint(0, MAX_IPV4))

 The DOGE guy just stripped the comments out.

Cool. I'd expect such a throwaway account to be a bare throwaway account and not have multiple learning-project style repos with activity spreading out over a period of a few years, such as you see with the author of the critique/rant posted on the DOGE guy's repo.

The original author claims this is to prevent API gateway from leaking the true client IP.

???

The public repos for this person that I could find that weren't forks with no activity to upstream consisted of a dice-rolling guessing game, rock-paper-scissors, and some kind of framework for downloading and transcribing audio files that does not yet download or transcribe, but implements a whole bunch of boilerplate. I find it rather difficult to believe this person engaged in a good-faith review of the async-ip-rotator code base.

I agree - I'm not endorsing this broad usage of the term, just noting how I see it used in practice. This seems to be half driven by grifters on social media promoting their AI coding course du jour and half by people who just discovered Cursor, Windsurf, etc., and think any use of those tools counts.)

Another commenter said these people only read half of Karpathy's tweet - I disagree. I don't think they read it at all, have heard the phrase, and are using it with reckless abandon.

The definition seems to have pretty rapidly moved to 'used an AI coding assistant in some capacity'.

This should be even more useful now that tire shops are refusing to repair tires with roadside plug repairs.

For b2b sales, as this IBM initiative seems to be focused on, I think there are a few things going on. This is all my opinion, based on first hand observation during the height of COVID WFH in enterprise sales, for a vendor that did very well during the lockdown period but which "returned to customer" as quickly as possible.

First, there was a lot of nervousness about long-term sales pipeline creation during lockdown. That anxiety was not completely unreasonable. While we had a lot of contact with our customers over video conferencing, etc., it was tactical and project-focused. The thought was we were getting locked out of all of the hallway conversations, lunches, conferences, trips, etc. where you tend to learn about new projects, problems that need to be solved, etc.

Second, sales leadership is a travel-heavy business. I spent 3 - 4 days a week, almost every week, on the road. That came to a rather abrupt halt in early 2020, which was fine with me. I never really liked the travel. But, as far as I could tell, I was in the minority in that belief. The job selects for the road warrior, and most of my peers and bosses could not wait to get back on the road.

And, so, I think people took a plausible hypothesis (pipeline will evaporate if we don't spend face time with our customers) that they wanted to be true, and ran with it.

At least as described by the OP, I have a hard time believing IBM is doing this. There are a whole bunch of markets -- far more than they listed in their flagship/hub site list -- where a vendor like IBM would have a several - a few dozen dedicated teams that just serve local clients. Now, in a lot of cases (but by no means all) the sales engineers can get badged at their customers and spend a few days a week onsite. While not unheard of, especially at the largest shops, its less common for the account managers to be badged and have unfettered access.

That said, I've never worked at IBM, and the sales segments they listed in the article (strategic, enterprise, etc.) are notoriously company-specific. But, I can't imagine the ones in the exception list cover all of teams with named accounts.

Deliberately placing your sales teams in different cities than their clients is almost always an incredibly dumb idea, WFH or not. In 15 years of sales engineering/sales engineering management, I remember exactly one account team we deliberately placed outside of their client geography, and that was because the client was so spread out, it made more sense to put the team close to a big airport than pick a single client site to place them near.

edit - I’m talking about dedicated/assigned teams, not inside sales or startups that have one or a handful of teams to cover everything.

She might well have gotten the idea there. This was a fairly dilbert-esque company. Somebody adapted the meeting with a vendor comic[0] with the company's name and a drawing of one of their products. It was used in a lot of internal (and some customer-facing) presentations.

0 - first one on this page https://kintronics.com/technical-jokes-cartoons-8/

I interviewed someone once for a network engineering job and, while talking about multicast routing, he mentioned Rendezvous Points[0], only he pronounced rendezvous as "Ron-divi-us". I asked him to describe RPs in a bit more detail (but pronounced rendezvous correctly[1],) and he said "oh, that's how you say that? That makes sense..." He had heard (and used) the phrase rendezvous point before - in the Army -- but didn't make the connection to the weird spelling he encountered with the multicast documentation.

0 - in Protocol Independent Multicast Sparse-mode routing, an RP is the root of the shared tree of participants for a given multicast group. See RFC 2362.

1 - for an English speaker, anyway. I imagine a native French speaker could pick apart the way we pronounce rendezvous.

(It is/was a subreddit where many observations of this nature are posted.)

...But it wasn't flagged. People just didn't keep upvoting it.

> If this was a top-down DOGE directive, why the immediate reversal? Did DOGE suddenly change its mind?

From the man himself:

  “We will make mistakes. We won’t be perfect. But when we make a mistake, we’ll fix it very quickly,” Musk, a Trump-appointed special government employee, said Wednesday in defense of his group’s haphazard cuts while looming over the Cabinet table. “So for example with USAID, one of the things we accidentally canceled—very briefly—was Ebola prevention."[0]

> The fact that the contract lasted 25 years _is_ relevant. It shows that this wasn't some minor side project. CVE is foundational infrastructure. You don’t accidentally let something like that expire unless someone either massively screwed up or there was serious internal confusion.

See previous quote about ebola funding, another long-term government program.

Look, if you want to play this game where because a DOGE spokesperson hasn't directly come out and said "yep, it was us", then I'll point you back to my original post: I said it was reasonable to assume, not that it was proven fact.

But, if you look at the totality of this situation and think 'nope. no way DOGE was involved. This is just people "blaming DOGE for everything"'. Fine. You do you. I don't think there is any point in continuing this conversation.

0 - https://newrepublic.com/post/192082/elon-musk-fact-check-dog...

What I'm doing is drawing a reasonable inference based on the evidence available to me. Specifically:

* The group is question (DOGE) is tasked with cutting spending deemed superfluous or wasteful.

* The group in question is actively cutting spending at the agency in question (CISA)

* The group in question is actively cutting spending with the vendor in question (MITRE)

* The leader of the group in question (Elon Musk) has said, more or less explicitly, that they have a bias more towards cutting spending and less towards getting the spending cuts right. They expect mistakes to be made. (If you want to dispute or nitpick this, I'll link you to the video where he laughs about cutting ebola prevention funding.)

* The boss of the group in question (Trump) really doesn't like CISA. See: the Chris Krebs debacle.

So, that's what I'm doing. What are you doing? It seems to me that you are trying to fit the available facts to your preferred narrative, instead of the other way around. And what does the contract having existed for 25 years have to do with anything? DOGE has admitted to cutting programs that have been funded for far longer. It is completely irrelevant.

This lede from NBC's article[0] last week covers it pretty well:

  President Donald Trump directed federal agencies Wednesday to revoke the security clearances and review the activities of two high-level former government officials who questioned his election fraud claims and his conduct in his first term.

  Christopher Krebs, the former head of the federal government's top cybersecurity agency, said Joe Biden won the 2020 elections, contradicting Trump's claim that the election was stolen.

Today, Krebs announced he was resigning from SentinelOne so as not to drag them in to what he described as his fight.

0 - https://www.nbcnews.com/politics/national-security/trump-ord...

The salient point is it was updated today with a statement from Krebs announcing his resignation.

Yes, imagining that the quasi-government organization that is solely tasked with cutting spending might have cut spending at an agency where they are currently cutting spending is a “huge leap.”

What was I thinking?

MITRE has been hit with DOGE-branded cuts[0] earlier this month. CISA has been impacted[1]. It seems reasonable to assume they were involved in this.

0 - https://virginiabusiness.com/nova-govcon-firm-mitre-to-lay-o...

1 - https://techcrunch.com/2025/03/11/doge-axes-cisa-red-team-st...

> Did CISA signal to MITRE that they weren't going to exercise the option?

An internal letter sent to CVE board members was making the rounds yesterday warning the current contract ("contracting pathway") would expire. The letter was authenticated by Brian Krebs[0]. Once Krebs authenticated the letter, people more or less assumed CISA was pulling funding, at least based on the infosec social media posts I saw.

CISA officials responded to multiple media inquiries (including the OP) with a statement that more directly said the contract would expire:

  Although CISA’s contract with the MITRE Corporation will lapse after April 16, we are urgently working to mitigate impact and to maintain CVE services on which global stakeholders rely.[1]

0 - https://krebsonsecurity.com/2025/04/funding-expires-for-key-...

1 - https://www.csoonline.com/article/3963190/cve-program-faces-...

?

Metacurity’s post was like 90 minutes ago.

Just social media posts, with claims they received the info from CISA https://infosec.exchange/@metacurity/114347467581760027

Supposedly, MITRE will make a statement today. Time will tell.

Edit - it is MITRE, not CISA, which the poster expects to make a statement.

The last one in my area shut down in 2014/2015. I stopped by and bought several large shopping bags worth of components, all for around $75. Now, I don't think there's anywhere locally to buy electronic components, except for industrial capacitors and things like that at trade supply shops.